structuring of surface for light management in monocrystalline si solar
TRANSCRIPT
STRUCTURING OF SURFACE FOR LIGHT MANAGEMENT IN
MONOCRYSTALLINE Si SOLAR CELLS
A THESIS SUBMITTED TO
THE GRADUATE SCHOOL OF NATURAL AND APPLIED SCIENCES
OF
MIDDLE EAST TECHNICAL UNIVERSITY
BY
SEDAT BİLGEN
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR
THE DEGREE OF MASTER OF SCIENCE
IN
PHYSICS
AUGUST 2015
Approval of the Thesis:
STRUCTURING OF SURFACE FOR LIGHT MANAGEMENT IN
MONOCRYSTALLINE Si SOLAR CELLS
submitted by SEDAT BİLGEN in partial fulfillment of the requirements for the
degree of Master of Science in Physics Department, Middle East Technical
University by,
Prof. Dr. Gülbin Dural Ünver
Dean, Graduate School of Natural and Applied Sciences
Prof. Dr. Mehmet Zeyrek
Head of Department, Physics
Prof. Dr. Raşit Turan
Supervisor, Physics Department, METU
Examining Committee Members:
Prof. Dr. Mehmet Parlak
Physics Department, METU
Prof. Dr. Raşit Turan
Physics Department, METU
Prof. Dr. Nizami Hasanli
Physics Department, METU
Assoc. Prof. Dr. Akın Bacıoğlu
Physics Engineering Department, HU
Assist. Prof. Dr. Selçuk Yerci
Micro and Nanotechnology Department, METU
Date:27.08.2015
iv
I hereby declare that all information in this document has been obtained and
presented in accordance with academic rules and ethical conduct. I also declare
that, as required by these rules and conduct, I have fully cited and referenced
all material and results that are not original to this work.
Name, Last Name: Sedat Bilgen
Signature :
v
ABSTRACT
STRUCTURING OF SURFACE FOR LIGHT MANAGEMENT IN
MONOCRYSTALLINE Si SOLAR CELLS
Bilgen, Sedat
MS, Department of Physics
Supervisor: Prof. Dr. Raşit Turan
August 2015, 108 pages
Texturing of a silicon wafer is the first process of production of screen
printed solar cells to reduce the reflection losses by producing pyramids on the
surface of the silicon wafer. Being a cheap and time efficient process, texturing is
used in all industrial applications. For mono-crystalline silicon wafers, the process is
carried out by using an alkaline solution which consists of potassium hydroxide
(KOH), isopropyl alcohol (IPA) and de-ionized water (DI-water) which is heated to
75- 80oC, and wafers are put in it up to a certain time to get random pyramids on the
surface. These pyramids reduce the reflection and increase light trapping, and thus
increase the efficiency.
In this study, the magnetic agitation was used to optimize the process
parameters like, the concentrations of KOH and IPA, process temperature, and
process time. To minimize the process duration and process temperature, ultrasonic
agitation was used, and almost uniformly distributed small pyramids (3 µm in
average) were obtained. To observe the surface characteristics, reflection
vi
measurements and SEM images were taken. Then, the simulations of uniformly
textured wafers with textured one side and two sides were made to observe the
absorption ability with different thicknesses and different pyramid heights. Then,
solar cells were fabricated by using new process parameters and agitation.
To characterize the solar cells, reflection, external quantum efficiency (EQE),
current-voltage (I-V), and Suns-Voc measurements were carried out. It was observed
that cells based on wafers textured with new parameters and ultrasonic agitation
could be used to reach higher conversion efficiency values compared to reference
cells.
Key Words: Texturing, KOH- IPA Solution, Ultrasonic and Magnetic Agitation,
Pyramids
vii
ÖZ
TEK KRİSTAL YAPILI SİLİSYUM GÜNEŞ HÜCRELERİNDEKİ PUL
YÜZEYİNİN IŞIK YÖNETİMİ İÇİN ŞEKİLLENDİRİLMESİ
Bilgen, Sedat
Yüksek Lisans, Fizik Bölümü
Tez Yöneticisi : Prof. Dr. Raşit Turan
Ağustos 2015, 108 sayfa
Yüzey şekillendirilmesi silikon pul üzerinde piramitler oluşturarak yansıma
kayıplarını azaltan güneş pili üretim aşamalarının birincisidir. Ucuz ve zaman
tasarruflu olması, yüzey şekillendirilmesi işlemini tüm endüstride kullanılır
yapmıştır. Tek kristal yapılı silikon pullar için bu işlem potasyum hidroksit (KOH),
izopropil alkol (IPA) ve iyonize su alkalin karışımının 75- 80oC ısıtılması ve pulların
yüzeyinde rastgele piramit oluşturmak için belli bir süre bu karışıma konulması ile
gerçekleştirilir. Bu piramitler yansımayı azaltır ve ışık hapsetmeyi, bu da verimliliği
arttırır.
Bu çalışmada, manyetik karıştırıcı kullanılarak KOH ve IPA konsantrasyonu,
işlem ısısı ve işlem süresi gibi işlem parametreleri eniyileştirildi. Bundan sonra,
işlem süresi ve işlem sıcaklığını azaltmak için ultrasonik karıştırıcı kullanıldı ve
hemen hemen homojen bir biçimde küçük piramitler (ortalama 3 µm ) elde edildi.
Yüzey karakteristiğini incelemek için yansıma ölçümü ve SEM görüntüleri alındı.
viii
Sonra, tek taraflı ve çift taraflı düzenli dağılmış piramitlere sahip pulların
simülasyonu, bu pulların soğurma kabiliyeti farklı kalınlıklarda ve farklı piramit
boyutlarında gözlemlemek için yapıldı. Yeni işlem parametreleri ile güneş hücreleri
üretildi.
Bu güneş hücrelerini karakterize etmek için yansıma, EQE, akım voltaj,
Suns-Voc ölçümleri yapıldı. Yeni parametrelerle ve ultrasonik karıştırma ile
yüzeyleri şekillendirilen pullardan üretilen hücrelerin referans olarak üretilen
hücreden daha yüksek verime sahip olduğu gözlemlendi.
Anahtar Kelimeler: Yüzey şekillendirilmesi, KOH-IPA solüsyonu, Ultrasonik ve
Manyetik Karıştırma, Piramitler
ix
To My Dearest Family
x
ACKNOWLEDGEMENTS
I would like to thank Professor Raşit Turan for his support through all steps
of this thesis and giving me the opportunity of working with him and his laboratory.
Thanks to him for being a member of GÜNAM family.
I would also like to thank to Fırat Es and Emine Hande Çiftpınar for helping
me to work in the clean room without killing myself and writing this thesis. I would
also like to thank to Mete Günöven, Olgu Demircioğlu and GÜNAM family.
Finally, I would like to thank my family and my wife for their patience and
support along my studies.
xi
TABLE OF CONTENTS
ABSTRACT ............................................................................................................ v
ÖZ ......................................................................................................................... vii
ACKNOWLEDGEMENTS ..................................................................................... x
TABLE OF CONTENTS ........................................................................................ xi
LIST OF FIGURES .............................................................................................. xiv
LIST OF TABLES ..................................................................................................xx
CHAPTERS
1. INTRODUCTION ............................................................................................... 1
1.1 History of Photovoltaic Technology ............................................................... 3
1.2 Types of Solar Cells ....................................................................................... 5
1.2.1 c-Si Wafer Based Solar Cells ................................................................... 5
1.2.2 Thin Film Solar Cells .............................................................................. 5
1.2.3 Other Solar Cell Types ............................................................................ 6
1.3 Basic Concepts of Photovoltaic Science and Technology ............................... 7
1.3.1 Solar Irradiation ....................................................................................... 7
1.3.2 Basics of Solar Cell Operations ............................................................... 9
2. FUNDAMENTALS OF CRYSTALLINE SILICON SOLAR CELL
TECHNOLOGY .....................................................................................................17
2.1 Introduction ...................................................................................................17
2.2 Texture: Wet Chemical Texture .....................................................................18
2.3 Doping ..........................................................................................................18
2.3.1 P-n Junction............................................................................................20
xii
2.3.2 Modeling of Solid State Diffusion .......................................................... 22
2.4 Anti Reflecting Coating (ARC) ..................................................................... 24
2.5 Metallization Screen Printing ........................................................................ 26
2.6 Edge Isolation ............................................................................................... 30
3. FUNDAMENTALS OF CHEMICAL ETCHING PROCESS AND LIGHT
TRAPPING ............................................................................................................ 33
3.1 Introduction .................................................................................................. 33
3.2 Wet Chemical Etching with KOH/IPA Solution ............................................ 36
3.3 Surface Morphology and Light Trapping ...................................................... 38
4. EXPERIMENTAL PROCEDURES.................................................................... 43
4.1 PROCESS FLOW OF SOLAR CELL FABRICATION ................................ 43
4.1.1 Texturing of Silicon Wafer ..................................................................... 43
4.1.2 Doping ................................................................................................... 44
4.1.3 Anti-Reflecting Coating ......................................................................... 45
4.1.4 Metallization and Co-Firing ................................................................... 46
4.1.5 Edge Isolation ........................................................................................ 48
4.2 Average Pyramid Heights and Pyramid Density, Mass Loss .......................... 49
4.3 Characterization of Solar Cells ...................................................................... 52
4.3.1 Reflection Measurements ....................................................................... 52
4.3.2 I-V Measurements .................................................................................. 54
4.3.3 Electroluminescence Measurements ....................................................... 54
4.3.4 External Quantum Efficiency (EQE) Measurements ............................... 55
4.3.5 Suns-Voc Measurements ........................................................................ 56
5. RESULTS AND DISCUSSION ......................................................................... 57
5.1 Texturing Process with Magnetic Agitation ................................................... 57
xiii
5.1.1 Effect of Process Duration ......................................................................57
5.1.2 Effect of KOH Concentration .................................................................62
5.1.3 Effect of using a cap to Keep IPA Constant ............................................69
5.1.4 Effect of Magnetic Stirring Speed ...........................................................72
5.1.5 Effect of IPA Concentration ...................................................................74
5.2 Texturing Process with Ultrasonic Agitation .................................................77
5.3 Simulation of Light Trapping ........................................................................83
5.3.1 Simulation of SDE wafer ........................................................................85
5.3.2 Simulation of Wafer with Uniformly Distributed Pyramids.....................86
5.4 Solar Cell Results ..........................................................................................91
CONCULUSIONS................................................................................................ 101
BIBLIOGRAPHY ................................................................................................. 105
xiv
LIST OF FIGURES
FIGURES
Figure 1: Percentage of different types of PV cells production in 2013 [2] ................ 2
Figure 2: Becquerel experimental setup to show the photovoltaic effect [3] .............. 3
Figure 3: Right is multi-crystalline solar cell. Left is mono-crystalline solar cell ....... 5
Figure 4: CIGS solar cells on flexible substrate [8] ................................................... 6
Figure 5: Organic Solar Cell [14] .............................................................................. 7
Figure 6: Solar Radiation Spectrum[3] ...................................................................... 8
Figure 7: The path length in units of Air Mass, changes with the zenith angle ........... 8
Figure 8: I-V curve of a Solar Cell. Blue dash line is under dark, and red solid line is
under illumination [16] ........................................................................................... 10
Figure 9: Typical I-V curve of a solar cell............................................................... 11
Figure 10: Circuit diagram of a solar cell including the shunt resistance [3] ............ 11
Figure 11: Variation of I-V curve with respect to shunt resistance Rsh. For ideal case
Rsh=10 kΩ cm2 [3] ................................................................................................. 12
Figure 12: Variation of I-V curve with respect to series resistance Rs [3] ................ 13
Figure 13: I-V curve of a solar cell with Rs=5.1 ohm cm2 [3] .................................. 14
Figure 14: Quantum Efficiency of a solar cell [3] ................................................... 15
Figure 15 : The production steps of screen printed crystal silicon solar cell ............. 17
Figure 16: The structure of a screen printed solar cell [6] ........................................ 18
Figure 17: (a) Crystal structure of intrinsic Si; (b) P doped Si crystal structure; (c) B
doped Si crystal structure ........................................................................................ 19
Figure 18: P type and N type semiconductors respectively [3] ................................ 21
Figure 19 : P-n junction [3] ..................................................................................... 21
Figure 20:(1) Energy band diagrams of a metal and an n-type semiconductor before
combined, (2) Band diagrams after combination of a metal and an n-type
semiconductor [25] ................................................................................................. 27
xv
Figure 21: :(1) Energy band diagrams of a metal and a p-type semiconductor before
combined, (2) Band diagrams after combination of a metal and a p-type
semiconductor [25] .................................................................................................28
Figure 22: I-V curve for Ohmic and Schottky contact .............................................29
Figure 23: :(1) Energy band diagrams of a metal and an n-type semiconductor before
combined, (2) Band diagrams after combination of a metal and an n-type
semiconductor for Ohmic contact [25] ....................................................................29
Figure 24: (1) Energy band diagrams of a metal and a p-type semiconductor before
combined, (2) Band diagrams after combination of a metal and a p-type
semiconductor for Ohmic contact [25] ....................................................................30
Figure 25: Solar cell operation before (1) and after (2) edge isolation [26] ..............31
Figure 26: AM 1.5 spectrum at global tilt [27] ........................................................33
Figure 27: Absorption depth of Si for 90% absorption [12] .....................................34
Figure 28: (a) The silicon crystal (b) Miller indices in a cubic structure...................35
Figure 29: Formation of square based pyramids of (100) oriented Si wafer .............39
Figure 30: Square pyramid with base length “a”, height “h” and half of the diagonal
of base “s”. The angle “α” between the pyramid edge and its projection “s” is equal
to “45o”
due to crystallographic structure [36]. ........................................................40
Figure 31: (a) Reflection from non-textured Si (b) Reflection from textured Si .......40
Figure 32: Ultrasonic cleaner ..................................................................................43
Figure 33: Hot plate and teflon holder .....................................................................44
Figure 34: Schematic of doping furnace [9] .............................................................45
Figure 35: The schematic representation of PECVD chamber [9] ............................46
Figure 36: ASYS EKRA-XL1 screen printing system ................................................47
Figure 37: Al mask for backside is in left, Ag mask for front side is in right............47
Figure 38: (a) Ideal case (b) Real case .....................................................................48
Figure 39: Finding number of pyramids by using the program ................................49
Figure 40: Wafer surface with uniform upright pyramids ........................................50
Figure 41: a) SEM image after converting black–white b) Area of black region ......51
Figure 42: Scaltec SBA31 balance ..........................................................................52
xvi
Figure 43: Reflection measurement setup [9] .......................................................... 53
Figure 44: Schematic presentation of solar simulator [9] ......................................... 54
Figure 45: Schematic representation of EQE measurement setup [9] ....................... 55
Figure 46: Reflection of textured as-cut wafers for process 5.4vol% IPA, 4.2wt%
KOH with magnetic agitation at (a) 70oC (b) 75
oC (c) 80
oC .................................... 58
Figure 47: SEM images of textured as-cut wafers with magnetic agitation with a
solution consisting of 5.4vol% IPA, 4.2wt% KOH for 15, 25, 35 and 45 min at 75 C
............................................................................................................................... 60
Figure 48: Reflection of textured SDE wafers for process 5.4vol% IPA, 4.2wt%
KOH with magnetic agitation at (a) 70o C (b) 75
o C (c) 80
o C ................................. 61
Figure 49: SEM images of textured SDE wafers with magnetic agitation with a
solution consisting of 5.4vol% IPA, 4.2wt% KOH for 15, 25, 35 and 45 min at 75 C
............................................................................................................................... 62
Figure 50: Reflection of textured as-cut wafers for process 5.4vol% IPA, 3.33wt%
KOH with magnetic agitation at (a) 70o C (b) 75
o C (c) 80
o C ................................. 63
Figure 51: SEM images of textured as-cut wafers with magnetic agitation with a
solution consisting of 5.4vol% IPA, 3.33wt% KOH for 15, 25, 35 and 45 min at 75 C
............................................................................................................................... 64
Figure 52: Area covered by pyramids with respect to time and KOH concentration
for AS-CUT wafers at 75oC .................................................................................... 65
Figure 53: SEM images of samples (a) textured with 5.4wt% KOH (b) textured with
3.33wt% KOH at 80oC ........................................................................................... 65
Figure 54: : Reflection of textured SDE wafers for process 5.4vol% IPA, 3.33wt%
KOH with magnetic agitation at (a) 70oC (b) 75
oC (c) 80
oC .................................... 66
Figure 55: SEM images of textured SDE wafers with magnetic agitation with a
solution consisting of 5.4vol% IPA, 3.33wt% KOH for 15, 25, 35 and 45 min at 75oC
............................................................................................................................... 67
Figure 56: Area covered by pyramids with respect to time and KOH concentration
for SDE wafers at 75oC(Lines are just added to guide the eye) ................................ 67
xvii
Figure 57: Area covered by pyramids of textured wafers with magnetic agitation with
a solution consisting of 5.4vol% IPA, 3.33wt% KOH for 15, 25, 35 and 45 min at (a)
70oC (b) 75
oC (c) 80
oC............................................................................................68
Figure 58: Reflection measurements of wafers textured with a solution consisting of
4.2wt % KOH, 5.4vol % IPA, 25 min at 75oC with and without a cap. ....................69
Figure 59: SEM images of as-cut wafers textured with a solution consisting of 5.4vol
% IPA, 4.2wt % KOH for 25 min process duration at 75oC a) without cap (coverage
21%) b) with cap (coverage 92%) ...........................................................................70
Figure 60: SEM images of SDE wafers textured with a solution consisting of 5.4vol
% IPA, 4.2wt % KOH for 25 min process duration at 75oC a) without cap (coverage
35%) b) with cap (coverage 82%) ...........................................................................70
Figure 61: Reflection measurements of wafers textured with a solution consisting of
4.2wt % KOH, 5.4vol % IPA, 45 min at 75oC with and without cap a) as-cut b) SDE
...............................................................................................................................71
Figure 62: SEM images of wafers textured with a solution consisting of 5.4vol %
IPA, 4.2wt % KOH for 45 min process duration at 75oC a) as-cut b) SDE with and
without using a cap .................................................................................................71
Figure 63: Reflection measurements of wafers textured with a solution consisting of
3.33wt % KOH, 5.4vol % IPA, 15min and 25 min at 70oC with and without cap a)
as-cut b) SDE ..........................................................................................................72
Figure 64: Reflection measurements for as-cut wafers textured with a solution
consisting of 3.33wt % KOH, 5.4vol % IPA, 15min and 25 min at 70oC with a speed
of 50 and 125 rpm ...................................................................................................73
Figure 65: Reflection measurements for SDE wafers textured with a solution
consisting of 3.33wt % KOH, 5.4vol % IPA, 25min and 35 min at 70 and 75oC with
a speed of 50 rmp and 125 rpm ...............................................................................74
Figure 66: Reflection values of as-cut wafers wafers textured with a solution
consisting of 3.33wt % KOH, different IPA concentrations a) 15min and b) 25 min at
70oC with a speed of 125 rpm .................................................................................75
xviii
Figure 67: Mass loss with respect to time and changing IPA concentration (Lines are
just added to guide the eyes) ................................................................................... 76
Figure 68: Textured wafer with a solution consisting of 40ml IPA, 3.33wt % KOH,
15 min at 70oC ........................................................................................................ 76
Figure 69: (a) Reflection values for textured as-cut wafers with magnetic agitation
(MA) and ultrasonic agitation (UA) (b) AM 1.5-weighted reflection (AM 1.5-WR)
values (c) Lowest reflection values obtained up to now(Lines are just added to guide
the eyes) ................................................................................................................. 78
Figure 70: SEM images of a) 75M_45min as-cut wafer b) 80K_45min SDE wafer
c) UA_15min as-cut wafer ...................................................................................... 79
Figure 71: Reflection measurements of as-cut wafers textured by ultrasonic agitation
at a) 70oC b) 60
oC c) AM 1.5 weighted reflection values (Lines are just added to
guide the eyes)........................................................................................................ 80
Figure 72: SEM images of as-cut wafers textured by ultrasonic agitation at a) 70oC b)
60oC ...................................................................................................................... 82
Figure 73: a) Pyramid density b) Average Pyramid height c) Mass loss for as-cut
wafers textured with ultrasonic agitation at 70 and 60oC(Lines are just added to guide
the eyes) ................................................................................................................. 83
Figure 74: Model of SDE wafer reflection simulation ............................................. 85
Figure 75: a) Reflection data b) Reflection c) Transmission d) Absorption with
respect to changing thickness (Lines are just added to guide the eyes) .................... 86
Figure 76: Simulation of textured wafer obtained by a) assembling one pyramid on a
rectangle volume b) cutting the a part of rectangle volume ..................................... 87
Figure 77: a) Uniformly textured wafer model b) Simulation model ....................... 87
Figure 78: a)Reflection values for model with 1,6,10 µm pyramids b) AM 1.5-
Weighted Reflection ............................................................................................... 88
Figure 79: Absorption graph of two side textured wafer with 6 µm pyramid height 89
Figure 80: Absorption of a) one side and two side textured wafer, b)one side wafer
with 1 µm pyramids, c) one side wafer with 6 µm pyramids d) one side wafer with
10 µm pyramids...................................................................................................... 90
xix
Figure 81: Model of a wafer with one side texturing ...............................................90
Figure 82: Graphs of Average a) Jsc b) Voc c) Fill factor e) Efficiency with respect
to texturing time d) Efficiency- Jsc, f) I-V curve of best cell ..................................93
Figure 83: Mass loss and Jsc with changing process durations a) at 60oC b) at 70
oC
.............................................................................................................................. .93
Figure 84: AM 1.5-weighted reflection and Average Jsc values with respect to
changing texturing time a) at 60oC b) 70
oC .............................................................94
Figure 85: Pseudo and real efficiencies and fill factors a),c) at 60oC b),d) at 70
oC
process temperature ................................................................................................95
Figure 86: a) J01 b) J02 with respect to texturing time ...............................................96
Figure 87: Electroluminescence images of a) Reference b) 70UA 45min solar cell..97
Figure 88: EQE and Reflection-IQE of solar cells produced by using wafers textured
for 45 min a)-c) at 60oC and b)-d) at 70
oC ...............................................................98
Figure 89: SEM image of a) 70UA 45min b) 60UA 45 min c) reference wafer after
texturing process .....................................................................................................99
xx
LIST OF TABLES
TABLES
Table 1: Parameters of three different AS-CUT/SDE wafer texturing processes using
chemical solutions based on KOH/IPA/DI-water for magnetic agitation. ................ 58
Table 2: Parameters of three different AS-CUT/SDE wafer texturing processes using
chemical solutions based on KOH/IPA/DI-water for magnetic agitation. ................ 62
Table 3: Parameters of two different AS-CUT/SDE wafer texturing processes using
chemical solutions to observe effect of using a cap ................................................. 69
Table 4: Parameters of the solutions to observe the effect of the magnetic stirring
speed ...................................................................................................................... 73
Table 5: Parameters of three different AS-CUT/SDE wafer texturing processes using
chemical solutions to observe effect of IPA concentration ...................................... 75
Table 6: Parameters of two different AS-CUT wafer texturing processes using
chemical solutions .................................................................................................. 79
Table 7: Absorption coefficient and refractive index values of mono-Si .................. 84
Table 8: Parameters of models ................................................................................ 89
Table 9: I-V measurement results by Solar Simulator ............................................. 91
Table 10: Suns-Voc measurement results ................................................................ 94
xxi
1
CHAPTER 1
INTRODUCTION
Energy is vital for human beings. The sources of energy, which are used
mostly today, are coal, natural gas, and petroleum. But these sources are limited (i.e.
not renewable), and they pollute air, water, and soil. These problems are forcing
people to find new energy sources. Some new energy sources such as solar, wind,
water, nuclear energy etc. have already been in use of people. However, some of
these sources suffer from problems such as high cost, limited availability, geological
difficulties, and environmental disturbances. For example, for a hydroelectrical
power application, there must be rivers and waterfalls available. In addition, in all
such power stains the natural environment is somewhat disturbed. Among the new
sources, solar energy is the most promising and most available energy source in the
world. Power coming from the Sun is about 174 petawatts (1 PW=1015
watts) and
51% of it is absorbed by oceans and land. In other words 89 PW can be used as an
energy source [1]. When converted to other energy types such as electricity, the solar
energy has the potential to provide all energy demand of the human being on our
planet. This conversion can be done in two ways: it can be converted to heat by
concentration and then used to generate steam for power generating turbines, or one
can use photovoltaic (PV) solar cells which convert the solar power to electricity
directly. The former one is called Solar Thermal Electricity (STE) which is outside
the scope of this study. The latter is the most widely used technology for the solar
energy conversion today. With the recent price reductions and improvements in the
performance, the PV industry has grown very rapidly in all around the world. At the
end of 2014, the total installed PV capacity exceeded 200 GWp, which is a
significant value proving the potential of PV technology in energy supply. Solar cells
which are the basic building elements of PV power applications comprises several
2
different technologies such as crystalline Si (c-Si, mono and multi), thin film
systems (a-Si, CdTe and CIGS). Among them c-Si wafer based cells are the most
commonly used solar cells in PV industry. Figure 1 compares the share of different
technologies in PV market today. From the technology and performance point of
view, we see enormous development in recent years. The conversion efficiency,
which is the major performance parameter for solar cells, has been improved with
intensive research and development activities. For wafers based Silicon (Si) solar
cells, efficiency of the cell has reached 26% which is very close to the upper
theoretical limit for these types of cells [2].
Figure 1: Percentage of different types of PV cells production in 2013 [2]
The cost of the generated electricity is a very important issue for choosing the
solar energy instead of the fossil fuels. To decrease cost per watt of solar cells, either
the cell efficiency should be increased or the cost of production should be reduced.
We observe R&D activities in both directions in many different locations around the
world. New device architecture and approaches are being designed and implemented
towards lower cost of energy generation [2]. Optimizations of the current production
steps and improving them have also been investigated intensively.
In this thesis, we focus on the texturization process of mono-crystal Si wafers,
which is the first step of producing a c-Si solar cell. The aim of texturization is to
create three dimensional structures on the surface of the wafer to reduce the
reflection, and thus improve the absorption of the solar radiation. However, forming
the best structure on a Si wafer surface with minimum material loss is still an issue
3
which needs to be addressed. By a proper optimization, one can obtain best
absorbing condition with minimum material loss. The aim of the study is to develop
a methodology to decrease the amount of removed Si, the temperature of solution,
the amount of chemical used for etching, and finally to obtain the lowest reflection
from the surface.
1.1 History of Photovoltaic Technology
Alexandre-Edmond Becquerel, a French physicist, discovered the
photovoltaic effect in 1839. Figure 2 shows his setup [3]. Two platinum (Pt)
electrodes were put in an acidic solution containing silver bromide AgBr or silver
chloride AgCl which are light sensitive material. When these light sensitive materials
are illuminated, they decompose into their ions, such as Ag+ and Cl
- ions, which are
then collected by the electrodes leading to electric current flow in the outside
circuitry.
Figure 2: Becquerel experimental setup to show the photovoltaic effect [3]
In 1876, by using solid selenium contacted with platinum, Adams and Day
observed photovoltaic effect. Then, Chapin, Fuller and Pearson used Si to produce
solar cell with an efficiency of 6% in Bell Laboratories in 1954 [4]. This was the first
4
reported solar cell produced using Si. During the development of Si solar cells, some
theoretical approaches were put forward. In 1961, the efficiency of Si solar cell being
a maximum point of 33% was found by William Shockley and Hans Queisser [5].
After that point, the development steps of production a Si solar cell can be sorted as;
In 1974, an etch solution consisting of sodium hydroxide (NaOH) was used to
texture the silicon surface, and the reflection losses due to a flat surface was
reduced [6].
In 1975, instead of using expensive vacuum metallization to produce front
Silver (Ag) and back Aluminum (Al) contacts, Ralph et al. used the screen
printing method used for the first time [6]. They produced antireflection
coatings by using titanium oxide and silicon dioxide. Their produced silicon
solar cells became the standard cell with the efficiency of about 10% in the
1970s [6].
In 1984, silicon nitride (SiNx) was used as an antireflective coating by
Kimura [6].
In 2006, 86% of wafer based solar cells were produced using screen printing
to form Ag front and Al back contacts, and chemical vapor deposition was
used to deposit SiNx as the antireflection coating on the front surface [6].
Today, the most commonly used metallization method in PV industry is the
screen printing method to produce Si solar cells, due to the low cost and practical
applications [7]. About 19% efficiencies have been obtained in industrial production
by improving the screen printing solar cells up to now.
There are three reasons why the Si solar cells are dominant in photovoltaic
industry. One of them is that Si is the second most abundant element in the Earth.
Second one is the electrical properties of Si. The electrical properties of Si are; a)
electron hole pairs can be generated under radiation, and b)Si can conduct these
electrons and/or holes, and this conductivity can be modified by introducing atoms of
other elements [6]. The last one is that Si has been used and studied for more than 40
years in semiconductor industry.
5
1.2 Types of Solar Cells
1.2.1 c-Si Wafer Based Solar Cells
Wafer based solar cells have 90% share of photovoltaic market today [2].
There are two types of wafers which are used to produce this type of solar cells. One
of them is the mono-crystalline silicon wafer. The entire wafer is in a single crystal
form edge to edge for mono-crystalline Si wafer. The other one is multi-crystalline
wafer which consists of different crystalline domains with different crystal
orientation. As can be seen from the Figure 3, one can easily distinguish which one is
mono or multi-crystalline solar cell.
Figure 3: Right is multi-crystalline solar cell. Left is mono-crystalline solar cell
Although mono-crystalline solar cells have higher efficiencies (about 24% in
lab environment), multi-crystalline solar cells, which have efficiency about 21% in
lab environment, are dominant in photovoltaic market [2]. The reason of this
situation is that multi-crystalline solar cells have lower production cost than mono-
crystalline solar cells have.
1.2.2 Thin Film Solar Cells
Thin film solar cells are produced by using a chemical or physical process to
deposit the active material on substrate. The substrate can be glass or another flexible
material as shown in Figure 4 [8]. The major thin film materials are Amorphous Si
6
(a-Si), Cupper Indium Gallium (di) Selenide (CIGS) and Cadmium Telluride (CdTe)
[9].
Figure 4: CIGS solar cells on flexible substrate [8]
By using a few micron thick material, very high absorption can be achieved.
This ability provides less material consumption during production of thin film solar
cells. But thin film solar cells have lower efficiency values compared to Si, and in
addition sunlight can spoil their forms and make them instable. These problems make
thin film solar cells less popular in photovoltaic technology when compared with c-
Si solar cell technology.
1.2.3 Other Solar Cell Types
Dye sensitized solar cells (DSSCs), tandem solar cells and organic solar cells
are examples of some other solar cell types. These new approaches are aiming to
reach low cost and/or high efficiency through non-vacuum production techniques, or
tandem structures trying to exceed the Shockley-Queisser limit. For DSSC and
organic cells, efficiencies as high as 10-11% have been demonstrated [10]. For
tandem cells, efficiencies up to 46 % have been achieved [2].
In tandem solar cells, III-V semiconductors like InAs, GaAs, InP are used and
they have the ability of absorbing different regions of solar radiation at different
layers of solar cells. This results in efficiencies up to 30.8% under 1 sun illumination
[11]. Due to the high production cost, tandem solar cells are used with systems to
7
concentrate the sun light to obtain high efficiency (about 40%) from small areas [2].
This is called Concentrated Photovoltaics (CPV).
For organic solar cells and DSSCs, their production cost is cheap, but they
have lowest efficiencies when compared with the others. Although they have low
efficiencies, their applications make them attractive. Organic solar cells can be
applied on large flexible substrates by a process based on roll to roll production
technique [12]. DSSC can be used as transparent solar cells over windows enabling a
more aesthetic and large surface integration in Building Integrated Photovoltaic
(BIPV) applications [13]. These cell types also suffer from the efficiency degradation
with time.
Figure 5: Organic Solar Cell [14]
1.3 Basic Concepts of Photovoltaic Science and Technology
1.3.1 Solar Irradiation
The sun generates huge amount of energy which is about 63 MW/m2 at its
surface. At just outside the Earth atmosphere, this energy is measured as 1360 W/m2
which is very small quantity compared to the energy at sun’s surface [15]. Due to the
atmosphere consisting water vapor, dust particles, gas molecules (carbon dioxide,
ozone), parts of the solar radiation passing through it is absorbed, reflected or
scattered. This leads to the solar intensity spectrum on the Earth’s as shown in Figure
6.
8
Figure 6: Solar Radiation Spectrum[3]
The Air Mass (A.M.) is defined as the ratio of path length of solar radiation through
the atmosphere to the shortest possible path length [3]. It is defined as;
Air Mass (A.M.) = 1/ cos(φ) (1)
Where φ is called zenith angle which is the angle between the sun and its vertical
axis.
Figure 7: The path length in units of Air Mass, changes with the zenith angle
9
Figure 7 shows A.M. 0, A.M. 1 and A.M. 1.5. A.M. 0 refers to the spectrum
just outside the atmosphere i.e. “zero atmosphere”. This quantity is used to
characterize solar cells which are used for space power applications. A.M. 1 refers to
the spectrum when the sun is at the vertical direction and A.M. 1.5 refers to the
spectrum when the sun makes an angle 48.2o with respect to the normal. Solar cells
used for commercial applications are characterized with respect to A.M. 1.5. This
optical path length results in a power of approximately 1000 W/m2
which is defined
as “1 SUN”.
1.3.2 Basics of Solar Cell Operations
A solar cell is a device which is produced by combining p-type and n-type
semiconductors like diodes. Its current-voltage characteristics are similar to diodes in
dark condition. This characteristic can be expressed by the following formula in
dark;
(2)
In the equation, I0 is reverse saturation current, q is electron charge, V is
voltage applied between the terminals of the cell, k is Boltzmann constant, T is the
cell temperature in Kelvin, and n is ideality factor which is 1 for ideal case. Under
illumination, Iop, which is the optically generated current, term is subtracted from the
Equation 2 and the equation becomes;
(3)
The Figure 8 shows I-V curve for a solar cell under dark and illumination condition.
10
Figure 8: I-V curve of a Solar Cell. Blue dash line is under dark, and red solid line is
under illumination [16]
To get the conventional I-V curve of solar cell, red curve is mirrored about horizontal
axis and the 1st quadrant is taken.
Short Circuit Current (Isc)
The short circuit current is defined as the point which I-V curve intersects the
I-axis. In other words, Isc is obtained when the applied voltage between the terminals
of solar cell is zero or equal. From the Equation 3, one can easily say that Isc equal to
Iop. This is the maximum current which can be generated by a solar cell.
Open Circuit Voltage (Voc)
The open circuit voltage is defined as the point which I-V curve intersects the
V-axis. In other words, the net flowing current through the solar cell is zero. From
Equation 3, Voc is the maximum voltage which is generated by a solar cell and can be
obtained as;
(4)
The Figure 9 illustrates the typical I-V curve of a solar cell.
11
Figure 9: Typical I-V curve of a solar cell
Shunt Resistance (Rsh)
Shunt resistance is one of the main reasons of the power loss in solar cells. It
is mainly due to easy current channels due to the defects across the junction region.
When Rsh is low, the generated current follows these current channels leading to
recombination, and thus reduction of the open circuit voltage across (Voc) the solar
cell. This causes a reduction in the generated power. This situation is illustrated in
the following figures; Figure 10 and Figure 11.
Figure 10: Circuit diagram of a solar cell including the shunt resistance [3]
12
Figure 11: Variation of I-V curve with respect to shunt resistance Rsh.
For ideal case Rsh=10 kΩ cm2 [3]
One can understand from the Figures that to get high power from a solar cell Rsh
should be as high as possible. Shunt resistance can be expressed as [3];
(5)
which is easily determined from the slope of the I-V at V=0.
Series Resistance (Rs)
There are three main factors which form the series resistance for a solar cell
namely: the movement of current through the emitter and base of the solar cell, the
contact resistance between the metal contact and the silicon, and the resistance of the
top and rear metal contacts [17]. As the Rs increases, the fill factor decreases, and Isc
also decreases for higher values of Rs as shown in the Figure 12.
13
Figure 12: Variation of I-V curve with respect to series resistance Rs [3]
For ideal case, the series resistance I should be zero. The series resistance can be
expressed as [3];
(6)
The series resistance is calculated from the slope of the I-V curve at V = Voc.
Fill Factor (FF)
Fill factor is the ratio of maximum power to the product of short circuit
current and open circuit voltage. The voltage at maximum power is shown as Vmp,
and the current at maximum power is shown as Imp. The fill factor is defined as;
(7)
The fill factor is also defined as the ratio of the area A to area B as shown in the
Figure 13.
14
Figure 13: I-V curve of a solar cell with Rs=5.1 ohm cm2 [3]
Conversion Efficiency
The efficiency of a solar cell is defined as the ratio of the maximum output
power which is maximum power as shown in Figure 13 to the input power, and it is
commonly used parameter to identify which cell is better than the other. The
efficiency is expressed as;
(8)
Here to define Pin, under which condition the efficiency is measured must be
considered. For commercial solar cell, A.M. 1.5 conditions are used. For this
condition, Pin is defined as the area of the solar cell multiplied by 1kW/m2. This
value is equal to 24.3 W for a 156 × 156 mm2 cell.
Quantum Efficiency (Q.E.)
Quantum efficiency is the ratio of the number of collected carriers by the
solar cell to the number of photons for a given energy incident on the solar cell [18].
15
If all incoming photons at a certain wavelength is absorbed, and all the generated
minority carriers are collected, the quantum efficiency at that wavelength will be 1 as
shown in the Figure 14.
Figure 14: Quantum Efficiency of a solar cell [3]
Recombination effects are responsible for the reduction in quantum efficiency
of a solar cell. For example blue light is absorbed near the surface, and if the front
surface passivation is not good, the quantum efficiency is reduced about these
wavelengths as shown in Figure 14. For silicon, silicon is transparent below 1.12 eV
which is the energy band gap of silicon and quantum efficiency is zero for
wavelengths beyond 1107 nm.
Spectral Response
Spectral response (S.R.) is a parameter which is the ratio of the current
generated by the solar cell to the power incident on the solar cell [3]. It is expressed
as;
(9)
16
17
CHAPTER 2
FUNDAMENTALS OF CRYSTALLINE SILICON SOLAR CELL
TECHNOLOGY
2.1 Introduction
Screen printing method is the most commonly used method to produce silicon
based solar cells by solar cell industry. The production steps are shown in Figure 15.
Figure 15 : The production steps of screen printed crystal silicon solar cell
Texturing is the first step of the production steps to form upright pyramids on
a Si-wafer, then a doping process is used to obtain p-n junction, after that an anti-
reflective coating which is a thin film of SiNx is made to passivate the emitter and
reduce the reflection. Then the screen printing method is used to deposit front and
back contacts, and electrical contacts between the Si wafer and the pastes are
established by a firing process. Finally, edge isolation is made to isolate the front
from the back side of the wafer, and characterization of the solar cell is conducted.
Figure 16 shows the structure of such a solar cell [6].
18
Figure 16: The structure of a screen printed solar cell [6]
2.2 Texture: Wet Chemical Texture
Texturing of a silicon wafer is the first process of production of screen printed
solar cells to reduce the reflection losses by producing pyramids on the surface of the
silicon wafer. The process is carried out by using an alkaline solution which consists
of potassium hydroxide (KOH) or sodium hydroxide (NaOH), isopropyl alcohol
(IPA) and de-ionized water (DI-water) which is heated to 75-80oC, and wafers are
put in it up to a certain time. This solution is known as KOH-IPA solution in the
photovoltaic industry.
The anisotropy of the etch solution is used to obtain pyramids on the wafer.
This anisotropic behavior of the solution is due to different crystal orientations
having different etch rates. KOH or NaOH etches (100) and (110) surfaces of the Si-
crystal with a higher rate with respect to (111) surface [19]. Coverage of the surface
with pyramids is obtained by optimizing temperature, time, concentrations of DI-
water, IPA and KOH or NaOH. In chapter 3, texturing process will be covered in
detail.
2.3 Doping
Doping is the second process of production of screen printed solar cells.
Before doping, a pre-cleaning step is applied for textured wafers in order to remove
all inorganic-organic contaminants and the natural oxide that may have grown on the
19
wafer surface by using DI-water, hydrofluoric acid (HF), hydrochloric acid (HCl)
and cleaned wafers are dried in hot air.
In general, doping is used to change the electrical properties of
semiconductors like conductivity, resistivity etc. by adding impurity atoms into
crystalline lattice. In solar cell industry, it is done by solid state diffusion process.
This process takes place into a tube shaped oven heated to 800–900oC to provide
enough energy in order to change a lattice atom with an impurity atom.
Silicon is an element of Group IV-A of periodic table and its atomic number
is 14. The Silicon electron configuration is 1s22s
22p
63s
23p
2. It means that a silicon
atom has to make four bonds to be stable as shown in Figure 17 a. When a V-A
group atom having five valance electrons (in solar industry, phosphorous (P) is used)
is added to obtain n type silicon, the P atom makes 4 bonds with the nearest Si atom
and the rest electron of it will be weakly bonded to it as shown in Figure 17 b. The
weekly bonded electron can be separated easily even at room temperature to move
through the lattice, and it makes the conductivity higher than before.
Figure 17: (a) Crystal structure of intrinsic Si; (b) P doped Si crystal structure; (c) B
doped Si crystal structure
An element of III-A group like Boron (B) having three valance electrons is
doped; there will be a missing electron which is equivalent to an additional hole, in
the structure as shown in Figure 17 c. More holes make the conductivity higher than
before. This type of material is called as p-type Si.
20
Standard solar cells are produced on p-type solar cells. N type layer is formed
by using liquid phosphorous oxychloride POCl3. POCl3 is converted into gas form by
nitrogen gas which flows through the liquid, and it is sent into a tube furnace at a
temperature about 840oC. During the flow of this mixture O2 gas is also sent to the
furnace, and the following reactions take place;
4POCl3(g) + 3O2(g) 2P2O5(g)+6Cl2 (10) [20]
2P2O5(g)+5Si(s)5SiO2(s)+4P(s) (11) [20]
xSiO2+yP2O5xSiO2.yP2O5 (12) [20]
In the first reaction, POCl3 reacts with O2, and P2O5 is formed. This reaction
occurs in predeposition step. Then, P2O5 reacts with Si, and SiO2 is formed on the
surface of the wafer, and P atoms are released as shown in the second reaction. P
atoms diffuse into Si crystal and replace Si atoms, and the p type region is converted
into n type. This reaction occurs in drive-in step. In the third reaction, P2O5 reacts
with SiO2, and the phosphorous silicate glass (PSG) is formed. After doping, PSG is
removed by using HF solution.
2.3.1 P-n Junction
As mentioned in the previous section, majority carriers are holes and
electrons, in p type and n type respectively. There are also ionized and fixed atoms in
crystal lattice which are negatively charged (like B-) in p type and positively charged
(like P+) in n type as shown in Figure 18.
21
Figure 18: P type and N type semiconductors respectively [3]
When p type and n type semiconductors are combined, electrons from the n-
type region diffuse to p type region and leave behind uncompensated positive ions.
Similarly, holes in p-type diffuse to n type and leave behind uncompensated negative
ions. The uncompensated ions generate an E-field, which prevents substantially the
diffusion of e- to p type and h
+ to n type in equilibrium condition. The Figure 19
illustrates that situation;
Figure 19 : P-n junction [3]
22
The region in which there are no free carriers is called depletion region. Also,
thanks to E field, the minority carriers (h+ for n type and e
- for p type) which are
close to the depletion region or generated within a diffusion length distance can drift
to the other side and be collected easily. And a "built in" potential Vbi due to E
field is formed at the junction.
2.3.2 Modeling of Solid State Diffusion
The doping process takes places by solid state diffusion of the dopant atoms.
This process is explained by Fick’s theory of diffusion which was derived by Adolf
Fick in 1855. The Fick’s first law is [21];
(13)
J is the diffusion flux which is defined as the amount of solute transferred per
unit area per unit time and has the units of mol/cm2.s.
D is diffusion coefficient or diffusivity of the material defining how easily an
atom could move throughout a known medium. Diffusivity defined with the
units of cm2/s , changes as a function of temperature and concentration.
C(x,t) is the concentration of atoms per unit volume of the medium and
defined with the units of mol/cm3.
is known as the concentration gradient which is the driving force of
the diffusion process.
The matter flows in the direction of decreasing solute concentration is stated
by using the negative sign in the eq. 13. In addition, spatial change in diffusion flux
must be equal to the change in solute concentration per unit time within a very small
volume of the system as imposed by the continuity. This situation is expressed in Eq.
14 [21];
(14)
23
Inserting eq. 13 into eq. 14, the Fick’s 2nd
law of Diffusion for one
dimensional flow is obtained as shown in eq. 15 [21];
(15)
Eq. 15 can be solved by using different boundary conditions. One of them is
that constant source concentration at the surface corresponding to the initial
condition C(x,0) = 0 and boundary conditions C(0,t)= Cs and C(∞ ,t) =0 leads to a
solution in the form of complementary error function as expressed in Eq. 16 below
where Cs is the surface concentration of dopant atoms [21].This situation formulize
the predeposition step of the diffusion process in which dopant including process gas
is fed to the system with a constant flow rate.
(16)
The drive in process takes place after predeposition step. In this process, a
thin layer of dopant atoms is already present on the wafer surface. For this process,
initial condition is C(x,0)=0 and the boundary conditions are ∞
and
C(∞, t) =0 , where QT is the total amount of dopant per unit area and will diffuse into
the Si. The resultant solution satisfying the initial and boundary conditions will be in
the form of Gaussian distribution function whose mathematical expression is given in
eq. 17 [21].
(17)
24
2.4 Anti Reflecting Coating (ARC)
The surface of the cell is usually covered by SiNx layer as anti reflection
coating. Deposition of the SiNx:H layer is done by using plasma-enhanced chemical
vapor deposition (PECVD). There are two significant reasons of deposing a SiNx:H
layer on the front side of the wafer. One of them is to reduce the reflection losses.
The reflection index of the SiNx:H layer is about 2 which is between that of Si (3.5)
and that of air ( approximately 1) for a wavelength of 600 nm. This makes the layer
working well as an antireflective coating. So, the wafer with SiNx has lower
reflection comparing with that of the bare wafer, and more light can be absorbed in
the solar cell [6]. The other reason is the passivation effect of the layer on the
emitter. The SiNx:H layer is very rich in hydrogen H, hydrogen also diffuses into the
Si-wafer and passivates some impurities or crystal defects [6].
As a passivation and antireflective layer, silicon dioxide (SiO2) can also be
used. However, use of SiO2 has two disadvantages. The first one is that the
processing temperature (about 1000oC) is high, and it does not passivate as efficient
as SiNx:H. The other one is the refraction index of SiO2 (1.5) is low.
Theory
Optical behavior of a surface is related to the dielectric constant. Dielectric
constant can be described as in eq. 18 [22];
(18)
In eq. 18, is dielectric constant, is real part of the refractive index, and
is the imaginary part of the refractive index, which is used to define absorption
coefficient ( )as shown in eq. 19 [22].
(19)
25
Reflectance for the light travelling from a medium with refractive index n0 to
a medium with refractive index ns at normal incidence is expressed as in eq. 20 [22];
(20)
If the material is coated with an ARC thin film with refractive index n1, the
eq. 20 becomes for the light of wavelength [22];
(21)
where is defined as the phase shift in the film with thickness d1 and can be
expressed as in eq. 22 [22];
(22)
θ1 is the angle between the light ray and the normal of the film surface.
When θ1 is zero (i.e. light at normal incidence), the minimum reflection is obtained at
which leads to eq. 23;
(23)
26
To minimize the reflection once more, the light reflected from the rear and
front sides of the thin film should be out of phase so interfere destructively [23]. To
get this situation, n1 can be chosen to satisfy the eq. 24 below [22].
(23)
By using the equations above, for a single wavelength the reflection can be
made zero. Since the peak power of the solar spectrum is close to wavelength 600
nm, refractive index and thickness of ARC thin film is chosen to minimize reflection
at this wavelength in photovoltaic industry [24]. At 600 nm, refractive index of Si is
ns= 3.939, air refractive index is n0=1. Therefore n1 can be chosen as approximately
2, and d1 can be chosen as 75 nm.
2.5 Metallization Screen Printing
Generated minority carriers in the depletion region and within the diffusion
length distance from the depletion region are collected by metal contacts which are
silver contacts on front and aluminum contacts on rear of the wafer. These contacts
are applied using screen printing method.
To get Ag metal in physical contacts with the front side of the cell, the paste
containing lead borosilicate glass (PbO-B2O3-SiO2) and organic compounds is used.
During the firing process, the lead borosilicate glass etches the ARC film (SiNx thin
film) and provides adhesion of Ag contacts. The firing process takes place at around
875oC .
For Al back contact, a similar process takes place. To overcompensate the
phosphorous doping on the backside and to form proper Back Surface Field (BSF),
the quantity of paste is very important [6].
The temperature of firing process must be adjusted to get good and optimal
solar cells. For temperatures below 800oC, good electrical contact between pastes
and silicon does not form, and this leads very high series resistance [6]. For
27
temperatures higher than 900oC, the pastes can diffuse too deeply into silicon and
cause short circuits between the front and back contacts [6].
When a metal and a semiconductor are combined, the electrical contact forms
in two ways; Schottky and Ohmic contact. These contacts are related to whether the
work function of metal is grater or less than the semi conductor.
Schottky Contact
When a metal and a semiconductor are combined, charge transfer occurs until
the Fermi levels get the same position. To understand this situation, we look energy
band diagrams shown in Figure 20.
Figure 20:(1) Energy band diagrams of a metal and an n-type semiconductor before
combined, (2) Band diagrams after combination of a metal and an n-type
semiconductor [25]
In the Figure above, is the electron affinity, Φm and Φs are work function
for metal and semiconductor respectively. Φm must be greater than Φs for
combination of metal and n-type semiconductor to get Schottky contact. As shown in
Figure 20 (2) conduction and valance band of semiconductor will bend downwards
28
to prevent further electron diffusion from semiconductor to metal after combination of
the materials. Schottky barrier (ΦB) is formed and electron flow is stopped. Due to
electron flow to metal, there are uncompensated donor ions, and these ions induce
negative charges at metal side as shown in the Figure 20.
Figure 21: :(1) Energy band diagrams of a metal and a p-type semiconductor before
combined, (2) Band diagrams after combination of a metal and a p-type semiconductor
[25]
Figure 21 shows the energy band diagrams for a metal and a p-type
semiconductor before and after the combination. To form Schottky barrier for a
metal and p-type semiconductor, Φm must be less than Φs. The bending of the
conduction/valance band is upward to prevent hole flow between two sides.
Ohmic Contact
For Ohmic contacts, there is no or negligible potential barrier being formed
by combination of a metal and a semiconductor. In this case, I-V characteristic is
linear and independent of the polarity of applied voltage [25] as shown in Figure 22.
29
Figure 22: I-V curve for Ohmic and Schottky contact
For ohmic contact, relation between work functions is reversed. In other
words, Φm must be less than Φs for n-type semiconductor-metal combination, and Φm
must be greater than Φs for p-type semiconductor-metal combination. Energy band
diagrams are shown for n-type and p-type in Figure 23 and Figure 24 respectively.
Figure 23: :(1) Energy band diagrams of a metal and an n-type semiconductor before
combined, (2) Band diagrams after combination of a metal and an n-type
semiconductor for Ohmic contact [25]
30
Figure 24: (1) Energy band diagrams of a metal and a p-type semiconductor before
combined, (2) Band diagrams after combination of a metal and a p-type semiconductor
for Ohmic contact [25]
Although current is allowed in one direction due to the potential barrier for
Schottky contact, current can flow in both directions for Ohmic contact. Therefore,
front and back contact of solar cells are required to be Ohmic. As mentioned above,
Ag is used for front side for n type region, and Al is used for back side for p type
region to get Ohmic contact.
2.6 Edge Isolation
During the doping process, front side, back side and edges of p-type silicon wafer is
doped to produce n-type region. Although Al is diffused and converts the n-type
region formed at backside into highly doped p-type (p+) region in metallization
process, edges are not converted. This situation forms possible paths for generated
minority carriers to be able to flow to p-side through the edges, instead of being
collected at an external load. Figure 25-1 on the next page illustrates this situation.
Edge isolation disconnects the electrical contact between the front and the back
contacts. This is accomplished by either plasma/chemical etching which removes n
31
regions on edges completely or by laser ablation which forms shallow grooves on the
solar cell. Figure 25-2 shows the operation of solar cell after edge isolation.
Finally, the solar cells current-voltage characteristics of solar cells are
obtained by measurements under AM1.5 spectrum illumination conditions in order to
characterize.
Figure 25: Solar cell operation before (1) and after (2) edge isolation [26]
32
33
CHAPTER 3
FUNDAMENTALS OF CHEMICAL ETCHING PROCESS AND
LIGHT TRAPPING
3.1 Introduction
Absorption of photon is the critical process for solar cell devices because
absorbed photons generate electron hole pairs which form the output power of the
solar cells. As mentioned above, the band gap of silicon is 1.12 eV, and wavelength
between 300-1100 nm is suitable for the photovoltaic conversion as shown in Figure
26.
Figure 26: AM 1.5 spectrum at global tilt [27]
34
The Figure 27 shows the thickness which is needed to absorb the incoming
wavelength. As can be seen from the Figure, wafers with 400 µm thickness are
needed to absorb 90 % of incoming intensity with wavelength between 300-1000 nm.
Figure 27: Absorption depth of Si for 90% absorption [12]
Use of thicker wafers causes problems like difficulties in collection of
generated electron-hole pairs having minority carrier diffusion lengths 14 µm for n
type and 140 µm for p type [23], heavy modules, high cost etc. In other words,
thicker wafers have high absorption but they have low collection probability whereas
thinner wafers have high collection probability whereas they have poor absorption at
long wavelengths. To avoid these problems, surface texturing is used.
Although there are several dry texturing process techniques reported in the
literature to produce low reflective textured surfaces, structures produced by dry
process like needle, tips, wires etc. are not suitable for reliable passivation of c-Si
surface in solar cells [28]. On the other hand, pyramids formed as a result of
different etch rates of (100) and (111) planes are of the orders of micrometer scale.
These properties of alkaline texture make it an essential step for the industrial mono-
c applications.
35
Figure 28 (a) shows the crystalline structure of Si which is a face centered
cubic structure with a base of two identical atoms located at the (0, 0, 0) and (1/4,
1/4, 1/4) position. Every silicon atom is located at the center of a tetrahedron and is
covalently bonded to four other atoms. The number of free bonds of silicon atoms is
the difference in surface planes which are represented by Miller indices as shown in
Figure 28 (b). Si atoms have one free bond in the (111) plane, which means Si atoms
in this plane are bounded to the Si crystal with three other Si atoms. On the other
hand, Si atoms in the other planes, like (100) and (110), are bounded with two other
Si atoms and have two free bonds. Therefore, energy needed to dissolve Si atoms in
(111) plane is higher than energy required to dissolve atoms in (100) and (110)
planes. The energy differences lead that texturing of Si wafer with KOH-IPA
solution which is an anisotropic etching process (i.e. etching rate is orientation
dependent). In the next section, explanation of the etching process is given [29].
(a)
(b)
Figure 28: (a) The silicon crystal (b) Miller indices in a cubic structure
36
3.2 Wet Chemical Etching with KOH/IPA Solution
Although there are many explanations [29]–[32] about the anisotropic and
selectivity effects of alkaline etch solutions on mono-crystalline silicon, the process
keeps its mystery, and further investigations should be done [6] . Among these
explanations, H. Seidel et al. [29] described well the mechanism of the etching
process [6]. In this section, the KOH based etching process is explained by using
Seidel explanations.
Two hydroxide ions OH- from the alkaline solution bind to Si atoms having
two free bonds on plane (100) and two electrons from OH- are injected in the
conduction band of the silicon as indicated in Equation 25.
(25)
The strength of Si-Si back bonds weakens due to the bonds between Si and
OH. A soluble silicon hydroxide (Si-OH) complex is obtained by breaking the Si-Si
back bonds of the Si(OH)2 which happens when two electrons from back bonds are
excited to the silicon conduction band. This is shown in Equation 26. The reactions
shown in Equations (25) & (26) occur more or less simultaneously.
(26)
The positive silicon hydroxide complex reacts with another two hydroxide
ions from the solution and orthosilicic acid Si(OH)4 is produced as shown in
Equation 27.
37
(27)
The Si(OH)4 can now get mixed into the solution by diffusion, and it is
unstable due to the high pH value greater than 12 of the solution. For these pH
values, the detachment of two protons from Si(OH)4 ,which forms the following
complex as shown in Equations 28 & 29, will occur. This silicate species was
observed by Raman spectroscopy measurements [31].
Si (OH) 4 Si O 2(OH) 2--
+ 2 H+ (28)
2 H+ + 2 OH
-2 H2O (29)
The electrons located near the surface from Equations (25) & (26) in the
conduction band can be transferred to the water molecules close to the surface of
silicon, thus hydroxide ions and hydrogen atoms are produced, and they recombine
to molecular hydrogen as shown in Equations (30) & (31).
4 H2O + 4 e- 4 H2O
- (30)
4 H2O- 4 OH
- + 4 H 4 OH
- + 2 H2 (31)
Si atoms on (111) plane have one free bond and react with one hydroxide ion as
shown in Equation (32).
(32)
Three back bonds of the surface silicon atom have to be broken. In analogy to
Equations (26) & (27), this will happen when three electrons are transferred to the
38
conduction band , and three OH- will bind to Si-OH complex as shown in Equations
(33) & (34). The reaction will continue as described above for a (100) plane.
(33)
[Si-OH]+++
+ 3 OH- Si(OH)4 (34)
According to Seidel study, there is no significant role of the potassium cations
K+ and IPA in the etching process of Si, and they can be neglected during forming
reaction equations [6].
The effects of IPA found in the literature in the etching process can be listed
below;
As the concentration of IPA increases, the etch rate decreases [33].
Its role is to move away H2 bubbles from surface by decreasing the silicon
surface tension and improving the wettability of the wafer surface [34], [35].
When this happen, the attraction between the etching solution and the surface
will form easily, and homogenous texturing will occur.
IPA adjusts the relative water concentration of the etchant [29]. It attracts
water concentration around itself, so it regulates the etching rate for solutions
with a constant KOH concentration and constant temperature [6].
3.3 Surface Morphology and Light Trapping
Due to the difference in the etching rates of different planes, square based
pyramids are formed during the etching process. The
Figure 29 shows the steps of formation of pyramids. A wafer with (100) orientation
is dipped into the KOH/IPA solution and etching starts. As mentioned above, (100)
39
plane is etched faster than (111) plane and the proportionality between them is 600/1.
The process continues with high etch rate until all surfaces become (111) planes.
After that point, the process will be very slow; i.e. no change will be observed.
Figure 29: Formation of square based pyramids of (100) oriented Si wafer
The pyramids, whose bases are formed by (100) planes and sides are formed
with (111) planes, will be formed on the surface of the mono-Si wafer at the end of
the process.
To observe surface morphology of textured wafers, scanning electron
microscope (SEM) is used, because SEM pictures have strong signal contrast, lateral
resolution and depth of sharpness [36]. By using SEM pictures, the edge of pyramids
can be measured and the height of the pyramids (pyramid size referred in the
literature) can be determined as shown in the Figure 30. As shown in the Figure, the
height of the pyramids is equal to “s” which is equal to half of the edge.
40
Figure 30: Square pyramid with base length “a”, height “h” and half of the diagonal of
base “s”. The angle “α” between the pyramid edge and its projection “s” is equal to
“45o”
due to crystallographic structure [36].
The formation of the pyramids on the wafer increases the optical path length
which is the distance being travelled by a photon without being absorbed. In other
words, texturing process makes silicon thicker optically without changing the actual
thickness of it and increases the probability of photon absorption. The Figure 31
illustrates this situation.
Figure 31: (a) Reflection from non-textured Si (b) Reflection from textured Si
41
As shown in the Figure 31 (a), the incoming light hits the surface and goes on
its path in the Si. It reaches the back side and leaves the Si. In this case huge amount
of light is reflected from the surface and leaves without being absorbed.
The Figure 31 (b) shows that incident light striking the textured surface. In
this case, some amount of incident light will be transmitted into wafers and some will
be reflected. Therefore, square of non-textured wafer reflection will be the limit one
can reach .The probability for more than one bounce of light depends on the angle
“Ø” (equal to 54.7o
in our case) which is the angle between Si surface (in our case
(100) plane) and the faces of the pyramids (in our case (111) plane) [37]. If the angle
“Ø” is less than 30o, the normally incident light will not receive more than one
bounce. If Ø is between 45o
and 60o, the double bounce of normally incident light
will occur. If Ø is greater than 60o, the triple bounce of light will occur [37]. When
the reflected light reaches the other pyramid, some amount of it will be again
reflected and transmitted. Therefore, the transmitted light into the wafer will be
increased and the reflection will be decreased. In the wafer, the light will not leave as
in the first case and it will be reflected more than one and the optical path will be
increased. As the optical path increases inside the wafer, more light will be absorbed
and more electron hole pair will be generated. In other words, the efficiency will
increase.
42
43
CHAPTER 4
EXPERIMENTAL PROCEDURES
4.1 PROCESS FLOW OF SOLAR CELL FABRICATION
4.1.1 Texturing of Silicon Wafer
Process flow of Si solar cell production starts with texturing process. An
ultrasonic bath shown in Figure 32 is used to agitate the KOH-IPA solution, and its
frequency is 35 kHz. It is capable of processing 25 wafers per run. Ultrasonic source
is used to get uniform temperature distribution and solution homogeneity during
reaction. The percentage of ingredients will be mentioned in chapter 5.
Figure 32: Ultrasonic cleaner
44
For magnetic stirring texturing process, a hot plate, teflon holders a glass
beaker (2.5 liter volume) shown in Figure 33 are used. The hot plate has a stirring
speed controller and temperature controller.
Figure 33: Hot plate and teflon holder
The process is followed by rinsing the textured wafers with DI-water,
cleaning in HF-HCL solution and drying them in hot nitrogen gas. While removing
metal impurities on the Si-wafer surface is done by HCl, the native oxide and the
impurities on surface are removed by HF which makes the wafer surface free of trace
metals [33].
4.1.2 Doping
Standard doping recipe optimized at GUNAM Laboratories is used to get 55
Ω/ sheet resistance. Doping process is done by using SEMCO Engineering
Incorporation MINILAB Series doping furnace. The dopant source is liquid
45
phosphorous oxychloride POCl3 transported into furnace by N2. Figure 34 shows the
schematic of doping furnace [9]. The process flow is mentioned in the section 2.3.
Figure 34: Schematic of doping furnace [9]
4.1.3 Anti-Reflecting Coating
By using SEMCO Engineering Incorporation MINILAB Series, a-SiNx
deposition is carried out in PECVD chamber. The chamber can take 10 wafers per
run. The schematic representation of the chamber is shown in Figure 35.
46
Figure 35: The schematic representation of PECVD chamber [9]
The chamber is filled with SiH4 and NH3 gases. The process is carried out at
380oC and 1 Torr. An RF generator, which works at 375 W with 50 kHz, is used to
obtain plasma. The chemical reaction of process is shown in Equation 35.
3SiH4(g) + 4NH3(g) Si3N4(s) + 12H2(g) (35)
4.1.4 Metallization and Co-Firing
ASYS EKRA-XL1 screen printing system has been used for metallization of
textured, doped, and coated with ARC Si-wafers. The system is in GÜNAM
laboratory and is shown in Figure 36.
47
Figure 36: ASYS EKRA-XL1 screen printing system
Backside is covered fully with Al, and front side is covered using a
metallization mask with Ag as mentioned in the section 2.5. The design of mask is
done by considering two issues. One of them is the shadowing effect. The busbar’s
and finger’s width should be as small as possible to obtain more interaction between
incoming light and front surface of the wafer. As mention above, more light means
more short circuit current. The design of back and front contacts are shown in Figure
37. The other one is to obtain lowest resistance, because the contact resistance
increases, the efficiency decreases.
Figure 37: Al mask for backside is in left, Ag mask for front side is in right
48
To overcome these two requirements, the width and height of the finger
shown in Figure 38 should be optimized. The ratio of the width and height is called
aspect ratio. The width of finger should be as small as possible to overcome
shadowing effect, and the height of finger should be as long as possible to get large
finger cross sectional area which leads less resistivity. The aspect ratio can be
between 0.20 and 0.30 in screen printing metallization [38].
Figure 38: (a) Ideal case (b) Real case
After finding the optimized aspect ratio and printing front contact and back
contact, the contacts are dried. Wafers with dried contacts are ready to co-firing
process by using a conveyor belt. During the process, the wafers are exposed to high
temperature under atmospheric condition. The temperature profile of the firing
furnace is like a sharp Gaussian distribution function with a peak temperature around
830oC. At the end of process, Al is melted and forms p
+ back surface field, and Ag
diffuses through the ARC layer and forms front contact without melting.
4.1.5 Edge Isolation
The last process for producing a Si-solar cell is the edge isolation. The
process is carried out by using an IR marking laser. The edges of wafer are scanned.
At the end of the process, 10-15 µm deep grooves are formed and the connection
between the back and front (p and n region) is interrupted.
49
4.2 Average Pyramid Heights and Pyramid Density, Mass Loss
By using SEM images, two parameters namely; average pyramid heights
(APH) and pyramid density (PD) can be found. To find these parameters, I used an
open source program “Imagej”. The program can identify the tips of the pyramids
and give the number of pyramids in the SEM images as shown in Figure 39. Also by
using the program, the area of the picture can be obtained. By using these quantities,
PD is equal to number of pyramids divided by area of image as shown in Equation
36.
Figure 39: Finding number of pyramids by using the program
(36)
To find the average pyramid height, it is assumed that the bases of pyramids
do not nest each other and the surface is covered with pyramids as shown in
50
Figure 40. As mentioned in section 3.3, the height of the pyramid is equal to “s”.
APH can be found by using the equation 37, 38 and 39.
Figure 40: Wafer surface with uniform upright pyramids
(37)
(38)
(39)
Using “ImageJ”, the area covered by pyramids can be estimated from the
SEM images. The image is converted to black-white image as shown in Figure 41(a)
51
where the black area is non-textured surface and the white area is surface with
pyramids. Then, the black-white image is analyzed by the program that gives area of
the black region as shown in Figure 41(b). By subtracting this quantity from 100, the
textured area can be obtained.
(a)
(b)
Figure 41: a) SEM image after converting black–white b) Area of black region
Mass loss of textured wafers is measured by using a balance, Scaltec SBA31
shown in Figure 42. The mass loss is calculated by measuring the weight of wafer
before and after texturing process and taking the ratio.
52
Figure 42: Scaltec SBA31 balance
4.3 Characterization of Solar Cells
4.3.1 Reflection Measurements
First characterization step is reflection measurement of textured wafers.
Reflection of a wafer shows the quality of the texturing process. In other words, as
the area without pyramids increases, the reflection increases. Also reflection of
wafers with ARC is measured.
The measurement setup is shown in Figure 43. The setup consists of an
integrating sphere, a lock-in amplifier, a chopper and controller, a calibrated silicon
detector, a halogen lamp, a monochromator, a focusing lens and a PC.
53
Figure 43: Reflection measurement setup [9]
Light coming from the halogen lamp is modulated by the chopper whose
frequency is not equal to multiples of 50 Hz to prevent the affects of network
electricity. Frequencies which are not equal to chopper’s frequency are filtered by the
connection between chopper controller and lock-in amplifier. After that, the spot of
modulated light is decreased by the diaphragm and focused on the wafer surface by
the converging lens. The reflected light is collected at the entrance slit of
monochromator after many reflections. The light passing through the
monochromator is detected by the Si-detector which is connected to the computer via
the lock-in amplifier. The reflection is measured between 350 nm and 1100 nm. To
calculate the reflection of the sample, the (total) reflectance of BaSO4 reference disk
that is used in (our regular) total reflectance measurements has been measured using
NIST traceable Spectralon Diffuse Reflectance Standard (WS-1-SL) from Labsphere
[39]. Therefore, the reflection of the sample will be equal to the ratio of the sample’s
intensity to that of calibration disk.
54
4.3.2 I-V Measurements
The measurement is carried out in GÜNAM laboratory by using AM1.5G
calibrated QUICK SUN-120CA-XL Solar Simulator which is shown schematically in
Figure 44. Simulator’s software provides cell efficiency, fill factor, series resistance,
shunt resistance, open circuit voltage, short circuit current and maximum power point
as output data.
Figure 44: Schematic presentation of solar simulator [9]
4.3.3 Electroluminescence Measurements
The measurement is carried out by using a MBJ closed box
Electroluminescence system in GÜNAM laboratory. The measurement is done by
applying bias to the cells to force them to radiate. No radiation will be observed in
dead regions where radiation does not occur and the generated picture gives us
information about cracks and contacts problems.
55
4.3.4 External Quantum Efficiency (EQE) Measurements
In this measurement, the number of electrons generated per incident photon
as a function of wavelength is determined. EQE measurement provides information
about the effects of contacts, the junction quality, and recombination dynamics
across the device. The setup is shown in Figure 45.
Figure 45: Schematic representation of EQE measurement setup [9]
The light is modulated by the chopper and passes through the
monochoromator. Then it is focused on the wafer between the fingers by the
converging lens. The generated electron hole pairs are collected by two probes which
are connected to the front Ag busbar and Al back surface. The resultant response is
monitored after the collected signal amplification by lock-in amplifier.
56
4.3.5 Suns-Voc Measurements
The measurement is carried out using Suns-Voc system in GÜNAM
laboratory. The results of the measurement give us information about the base
material and passivation quality. In addition, the I-V behavior of the cell can be
predicted by analyzing the obtained data under zero series resistance assumption
which leads to estimate some important cell parameters like pseudo efficiency,
pseudo fill factor etc., because the effect of shunt and series resistance could have
been separated from each other. Power loses due to series resistance can be estimated
by comparing efficiency measurement results obtained by using solar simulator and
Suns-Voc [40].
57
CHAPTER 5
RESULTS AND DISCUSSION
5.1 Texturing Process with Magnetic Agitation
In the first part of the study, the effects of temperature, time, and
concentration of the ingredients were investigated by using the hot plate. In addition,
the effect of surface morphology was investigated by using as-cut and saw damage
etched (SDE) wafers. In this part, 5x5 cm2, p-type, 1-3 Ω-cm, (100) CZ-Si wafers
were textured. After texturing process, reflection measurements and SEM images
were taken to obtain texture quality and surface morphology.
5.1.1 Effect of Process Duration
To observe the effect of process duration, the etching solution consisting of
1500 ml DI-water, 63 g KOH, and 80 ml IPA was prepared. The etching duration
was up to 45 minutes. The temperatures were 70, 75, and 80oC. Table 1 shows the
parameters of six different as-cut and SDE wafer texturing processes using chemical
solutions based on KOH/IPA/DI-water for magnetic agitation.
58
Table 1: Parameters of three different AS-CUT/SDE wafer texturing processes using
chemical solutions based on KOH/IPA/DI-water for magnetic agitation.
Process
Name
IPA
(vol%)
KOH
(wt%)
Temperature
(C)
Agitatio
n rate
(rpm)
Process
Time
(min)
70M 5.4 4.2 70 50 15,25, 35,45
75M 5.4 4.2 75 50 15,25,
35,45
80M 5.4 4.2 80 50 15,25,
35,45
Figure 46 shows the reflection measurement results for as-cut wafers. As can
be seen from the figure, as the process time increases, the reflection decreases. The
minimum reflection is obtained for 45 min texturing. When the reflection results at
three different temperatures are compared, the minimum reflection is obtained in
process 75M. Therefore, the optimum temperature for these concentrations is 75o C
for as-cut wafers being textured with magnetic agitation.
300 400 500 600 700 800 900 1000 1100 1200
10
20
30
40
50
60
70
80
90
Wavelength (nm)
non-textured
non-textured^2
70M_15min
70M_25min
70M_35min
70M_45min
Re
fle
cti
on
(%
)
(a)
300 400 500 600 700 800 900 1000 1100 1200
10
20
30
40
50
60
non-textured
non-textured^2
75M_15min
75M_25min
75M_35min
75M_45min
Re
fle
cti
on
(%
)
Wavelenght (nm)
(b)
Figure 46: Reflection of textured as-cut wafers for process 5.4vol% IPA, 4.2wt% KOH
with magnetic agitation at (a) 70oC (b) 75
oC (c) 80
oC
59
300 400 500 600 700 800 900 1000 1100 1200
10
20
30
40
50 non-textured
non-textured^2
80M_15min
80M_25min
80M_35min
80M_45min
Re
fle
cti
on
(%
)
Wavelength (nm)
(c)
Figure 46: (Continued)
Figure 47 shows SEM images of as-cut wafers for process 5.4vol% IPA,
4.2wt% KOH at 75oC. As can be seen from the images, as the texturing time
increases, area with pyramids increases which leads to lower reflection. The pyramid
heights can be identify clearly, and they are between 1.5 -11 µm for 15 min
texturing, 3 -16 µm for 25 min texturing, 9 -20 µm for 35 min texturing. After 35
min texturing, the bases of pyramids nest and it makes difficult to identify the heights
correctly. It can be said that as the process duration increases, the pyramid heights
also increase.
60
Figure 47: SEM images of textured as-cut wafers with magnetic agitation with a
solution consisting of 5.4vol% IPA, 4.2wt% KOH for 15, 25, 35 and 45 min at 75 C
Figure 48 shows the reflection measurement results for SDE wafers. The
results show the same trend as that of as-cut wafers. As the process duration
increases, reflection decreases. Again the minimum reflection is obtained by 75o C
45 min texturing and it is almost equal to reflection values of as-cut wafer with
respective wavelengths.
61
300 400 500 600 700 800 900 1000 1100 1200
10
20
30
40
50
60
70
80
Wavelength (nm)
non-textured
non-textured^2
70M_15min
70M_25min
70M_35min
70M_45min
Re
fle
cti
on
(%
)
(a)
300 400 500 600 700 800 900 1000 1100 1200
10
20
30
40
50
60
70
80
Wavelength (nm)
non-textured
non-textured^2
75M_15min
75M_25min
75M_35min
75M_45min
Re
fle
cti
on
(%
)
(b)
300 400 500 600 700 800 900 1000 1100 1200
10
20
30
40
50
60
70
Wavelength (nm)
non-textured
non-textured^2
80M_15min
80M_25min
80M_35min
80M_45min
Re
fle
cti
on
(%
)
(c)
Figure 48: Reflection of textured SDE wafers for process 5.4vol% IPA, 4.2wt% KOH
with magnetic agitation at (a) 70o C (b) 75
o C (c) 80
o C
Figure 49 shows the SEM images of SDE wafers textured at 75o C. Again the
area without pyramid decreases by increasing process duration. The pyramid heights
are 3 -10 µm for 15 min texturing, 4.5-16 µm for 25 min texturing, 5 -20 µm for 35
min texturing. The pyramid heights also increase with process duration.
62
Figure 49: SEM images of textured SDE wafers with magnetic agitation with a solution
consisting of 5.4vol% IPA, 4.2wt% KOH for 15, 25, 35 and 45 min at 75 C
5.1.2 Effect of KOH Concentration
To see the effect of KOH concentration, wafers were textured by using
concentrations given in Table 2. In these processes, KOH concentration was
decreased to 3.33 wt%, and the other parameters were the same as indicated in Table
1.
Table 2: Parameters of three different AS-CUT/SDE wafer texturing processes using
chemical solutions based on KOH/IPA/DI-water for magnetic agitation.
Process
Name
IPA
(vol%)
KOH
(wt%)
Temperature
(C)
Agitation
rate
(rpm)
Process
Time
(min)
70K 5.4 3.33 70 50 15,25, 35,45
75K 5.4 3.33 75 50 15,25,
35,45
80K 5.4 3.33 80 50 15,25,
35,45
63
Figure 50 shows the reflection measurements of the as-cut wafers. At 70o C, it
can be seen that the reflection of 70K process is lower than that of 70M process for
all process durations when the Figure 48 (a) and Figure 50 (a) are compared.
At 75o C, when the Figure 48 (b) and Figure 50 (b) are compared, reflection for 75K
is lower than that of 75 M process for 15, 25, and 35 min process durations. The
reflection value for 75K is larger than that of 75M for 45 min process duration, but
the difference is almost 1%. At 80o
C, the situation is the same as 70K. Therefore,
decreasing the KOH concentration to 3.33wt % makes the magnetic agitation more
effective for shorter process durations than 45 min.
300 400 500 600 700 800 900 1000 1100 1200
10
20
30
40
50
Wavelength (nm)
non-textured
non-textured^2
70K_15min
70K_25min
70K_35min
70K_45min
Re
fle
cti
on
(%
)
(a)
300 400 500 600 700 800 900 1000 1100 1200
10
20
30
40
50
Wavelength (nm)
non-textured
non-textured^2
75K_15min
75K_25min
75K_35min
75K_45minR
efl
ec
tio
n (
%)
(b)
300 400 500 600 700 800 900 1000 1100 1200
10
20
30
40
50
Wavelength (nm)
non-textured
non-textured^2
80K_15min
80K_25min
80K_35min
80K_45min
Re
fle
cti
on
(%
)
(c)
Figure 50: Reflection of textured as-cut wafers for process 5.4vol% IPA, 3.33wt%
KOH with magnetic agitation at (a) 70o C (b) 75
o C (c) 80
o C
64
Figure 51 shows the SEM images for process 75K. When pyramids heights
are compared, 75K pyramid heights are smaller than that of 75M. As process
duration increases, area without pyramids decreases as shown in Figure 52 showing
area covered by pyramids. From Figure 52 , the area filled by pyramids for 75K
process is larger than that of 75M for 15 min, 25 min and 35 min process durations.
Figure 51: SEM images of textured as-cut wafers with magnetic agitation with a
solution consisting of 5.4vol% IPA, 3.33wt% KOH for 15, 25, 35 and 45 min at 75 C
65
15 20 25 30 35 40 45
20
40
60
80
100
Are
a c
ov
ere
d b
y p
yra
mid
s (
%)
Process Duration (min)
75M_Ascut
75K_Ascut
Figure 52: Area covered by pyramids with respect to time and KOH concentration for
AS-CUT wafers at 75oC
Figure 53 shows the images of 80M and 80K for 45 min process duration. As
can be seen from the figure, although the wafer textured with 4.2wt% KOH is not
covered with pyramids fully, the whole surface of the wafer textured with 3.33wt%
KOH is covered. From the SEM images, it can be concluded that increasing process
duration or temperature is not enough to get fully covered surface, i.e. all
components of the etch solution must be optimized.
Figure 53: SEM images of samples (a) textured with 5.4wt% KOH (b) textured with
3.33wt% KOH at 80oC
66
Figure 54 shows the reflection measurements for SDE wafers. When the
results are compared, the wafers textured with 4.2wt% KOH have again larger
reflection values than the wafers textured with 3.33wt% KOH up to 10% for 15 min,
25 min, and 35 min process durations. For 45 min process duration, the differences
are about 2%, 1%, and 5% at 70, 75, and 80oC respectively.
When Figure 50 and Figure 53 are compared, the reflection for as-cut and
SDE wafers are different from each other about %2 in average at same process
temperature and same process time. It showed that the process with these parameters
was less sensitive to surface morphology than the process parameters with 4.2wt%
KOH.
300 400 500 600 700 800 900 1000 1100 1200
10
20
30
40
50
60
Wavelength (nm)
non-textured
non-textured^2
70K_15min
70K_25min
70K_35min
70K_45min
Re
fle
cti
on
(%
)
(a)
300 400 500 600 700 800 900 1000 1100 1200
10
20
30
40
50
60
Wavelength (nm)
non-textured
non-textured^2
75K_15min
75K_25min
75K_35min
75K_45min
Re
fle
cti
on
(%
)
(b)
300 400 500 600 700 800 900 1000 1100 1200
10
20
30
40
50
60
Wavelength (nm)
non-textured
non-textured^2
80K_15min
80K_25min
80K_35min
80K_45min
Re
fle
cti
on
(%
)
(c)
Figure 54: : Reflection of textured SDE wafers for process 5.4vol% IPA, 3.33wt%
KOH with magnetic agitation at (a) 70oC (b) 75
oC (c) 80
oC
67
Figure 55 and Figure 56 show the SEM images of SDE wafers textured at
75oC and the graph of covered area with respect to time respectively. The wafers
textured with 3.33wt % KOH have larger area with pyramids than wafers textured
with 4.2wt % KOH have for process durations up to 35 min as can be seen from the
figures. For 45 min process time, both are covered fully with pyramids.
Figure 55: SEM images of textured SDE wafers with magnetic agitation with a solution
consisting of 5.4vol% IPA, 3.33wt% KOH for 15, 25, 35 and 45 min at 75oC
15 20 25 30 35 40 45
20
40
60
80
100
4.2wt% KOH
3.3wt% KOH
Are
a c
ov
ere
d b
y p
yra
mid
s (
%)
Process Duration (min)
Figure 56: Area covered by pyramids with respect to time and KOH concentration for
SDE wafers at 75oC(Lines are just added to guide the eye)
68
It is realized that both SDE and as-cut wafers textured with magnetic
agitation with a solution consisting of 5.4vol% IPA, 3.33wt% KOH have low
reflection and large coverage with pyramids for process durations 15 min, 25 min,
and 35 min. Figure 57 shows the area covered by pyramids with respect to different
process durations and temperatures. As can be seen from the Figure 57, process does
not depend on surface morphology very much for the solution with 3.3wt% KOH
and the minimum coverage is about 80% for 15min process duration. For the
following experiments, KOH concentration was kept constant as 3.33wt%, and the
other parameters were changed to get full covered wafers with pyramids.
Figure 57: Area covered by pyramids of textured wafers with magnetic agitation with a
solution consisting of 5.4vol% IPA, 3.33wt% KOH for 15, 25, 35 and 45 min at (a) 70oC
(b) 75oC (c) 80
oC
69
5.1.3 Effect of using a cap to Keep IPA Constant
Closing the beaker increases the pressure on the solution, and the evaporation
of water and IPA decreases. This makes the concentrations almost constant during
process. To see this effect on process, wafers were textured with parameters given in
Table 3.
Table 3: Parameters of two different AS-CUT/SDE wafer texturing processes using
chemical solutions to observe effect of using a cap
Process
Name
IPA
(vol%)
KOH
(wt%)
Temperature
(C)
Agitation
rate (rpm)
Process Time
(min)
75MC 5.4 4.2 75 50 25,45
70KC 5.4 3.33 70 50 15,25
Figure 58, Figure 59 and Figure 60 show the reflection results, SEM images
of as-cut wafers and SEM images of SDE wafers for processes 75M and 75MC. As
can be seen from Figure 58, reflection decreases dramatically for as-cut and SDE
wafers, but they are still high. Figure 59 and Figure 60 indicate that coverage with
pyramids increases from 21% to 92% for as-cut wafer and from 35% to 82% for SDE
wafer. The results show that using cap can decrease reflection or increases coverage
for process 75M_25min.
300 400 500 600 700 800 900 1000 1100 120010
15
20
25
30
35
40
45
50
55
60
Re
fle
cti
on
(%
)
Wavelength (nm)
75M_ascut
75MC_ascut
75M_sde
75MC_sde
Figure 58: Reflection measurements of wafers textured with a solution consisting of
4.2wt % KOH, 5.4vol % IPA, 25 min at 75oC with and without a cap.
70
Figure 59: SEM images of as-cut wafers textured with a solution consisting of 5.4vol %
IPA, 4.2wt % KOH for 25 min process duration at 75oC a) without cap (coverage 21%)
b) with cap (coverage 92%)
Figure 60: SEM images of SDE wafers textured with a solution consisting of 5.4vol %
IPA, 4.2wt % KOH for 25 min process duration at 75oC a) without cap (coverage 35%)
b) with cap (coverage 82%)
Figure 61 and Figure 62 show reflection measurements and SEM images of
wafers with a solution consisting of 4.2wt % KOH, 5.4vol % IPA, 45 min at 75 C
with and without a cap. From Figure 61, using a cap decreases the AM 1.5 weighted
reflection about 0.5% for as-cut and 0.8% for SDE wafers. When the images shown
in Figure 62 are analyzed, the average pyramid heights and density are 6.74 µm and
1.12x104 mm
-2 for as-cut, 7.5 µm and 0.889x10
4 mm
-2 for SDE wafers textured
71
without using a cap. Using a cap decreases the pyramid height to 2.38 µm for SDE
and to 2.19 µm for as-cut, and increases pyramid density to 0.883x105 mm
-2 for SDE
and to 1.04x105 mm
-2 for as-cut.
500 600 700 800 900 10009
10
11
12
13
14
15
Re
fle
cti
on
(%
)
Wavelenght (nm)
75M_45min_ascut
75MC_45min_ascut
12.9%12.4%
(a)
500 600 700 800 900 10008
9
10
11
12
13
14
15
Wavelength (nm)
Re
fle
cti
on
(%
)
75M_45min_SDE
75MC_45min_SDE
12.9%
12.1%
(b)
Figure 61: Reflection measurements of wafers textured with a solution consisting of
4.2wt % KOH, 5.4vol % IPA, 45 min at 75oC with and without cap a) as-cut b) SDE
Figure 62: SEM images of wafers textured with a solution consisting of 5.4vol % IPA,
4.2wt % KOH for 45 min process duration at 75oC a) as-cut b) SDE with and without
using a cap
72
Figure 63 shows the reflection measurements of wafers textured with a
solution consisting of 3.33wt % KOH, 5.4vol % IPA, 15min and 25 min at 70 C with
and without cap. Using a cap increases the reflection very much. Therefore, SEM
images of this process were not taken.
These results show that only using a cap is not enough to get fully covered
surfaces for short process durations (i.e. like 15 min and 25 min), and small pyramids
shows lower reflection than big ones although the difference is not very much. After
that point,effects of increasing the speed of magnetic stirring were observed.
300 400 500 600 700 800 900 1000 1100 120010
15
20
25
30
35
40
45
50
55
60
65
70KC_15min_ascut
70KC_25min_ascut
70K_15 min_ascut
70K_25 min_ascut
Wavelength (nm)
Re
fle
cti
on
(%
)
(a)
300 400 500 600 700 800 900 1000 1100 1200
15
20
25
30
35
40
45
50
55
Re
fle
cti
on
(%)
Wavelength (nm)
70KC_15min_sde
70KC_25min_sde
70K_15 min_sde
70K_25 min_sde
(b)
Figure 63: Reflection measurements of wafers textured with a solution consisting of
3.33wt % KOH, 5.4vol % IPA, 15min and 25 min at 70oC with and without cap a) as-
cut b) SDE
5.1.4 Effect of Magnetic Stirring Speed
To see the effect of magnetic stirring speed, wafers were textured with
parameters given in Table 4. These parameters were chosen because previous
process 70KC whose parameters given in Table 3 gives high reflection values. The
aim was to see whether there would be a recovery in texturing or not.
73
Table 4: Parameters of the solutions to observe the effect of the magnetic stirring speed
Process
Name
IPA
(vol%)/ ml
KOH
(wt%)
Temperature
(C)
Agitation
rate (rpm)
Process Time
(min)-Wafer
Surface
70KCS 5.4 3.33 70 125 15,25-
ASCUT
25,35-SDE
75KCS 5.4 3.33 70 125 25,35-SDE
Figure 64 shows the reflection results for process 70KCS as-cut wafers.
When the agitation speed was increased to 125 rpm, the reflection decreased. The
reflection and the coverage are related as the reflection decreases, coverage
increases, so coverage also increases with increasing speed. But the reflection value
was higher than the desired one about 10% for 15min and 5% for 25min and
wavelengths between 550 and 1000 nm. Therefore I did not put the SEM images of
these wafers.
300 400 500 600 700 800 900 1000 1100 1200
15
20
25
30
35
40
45
50
55
60
65
Re
fle
cti
on
(%
)
Wavelength (nm)
70KC_15min
70KC_25min
70KCS_15min
70KCS_25min
Figure 64: Reflection measurements for as-cut wafers textured with a solution
consisting of 3.33wt % KOH, 5.4vol % IPA, 15min and 25 min at 70oC with a speed of
50 and 125 rpm
74
Figure 65 shows the reflection measurements for SDE wafers textured with
solution parameters given above. At 70oC, SDE wafers showed high reflection values
for process durations 25min and 35 min, although the process was carried out by
using cap and with a high agitation speed. Process was repeated at 75oC with a high
agitation speed, and the reflection values decreased to sublimit which is square of
reflection of non-texture SDE wafer. Therefore, to get fully textured SDE wafers at
short process time with using a cap, minimum temperature should be 75oC.
300 400 500 600 700 800 900 1000 1100 1200
10
15
20
25
30
35
40
45
50
70KCS_25min
70KCS_35min
75KCS_25min
75KCS_35min
70K_25min
70K_35min
non-texture^2
Re
fle
cti
on
(%
)
Wavelength (nm)
Figure 65: Reflection measurements for SDE wafers textured with a solution consisting
of 3.33wt % KOH, 5.4vol % IPA, 25min and 35 min at 70 and 75oC with a speed of 50
rmp and 125 rpm
5.1.5 Effect of IPA Concentration
To see the effect of IPA concentration, wafers were textured with parameters
given in Table 5. In this experiment, the parameters were 70oC, 3.33wt % KOH, 125
rpm agitation speed, process durations 15 min and 25 min like process parameters of
70K. The differences from 70K were using a cap and the agitation speed. The IPA
was added in three different volume 40 ml, 80 ml, and 120 ml.
75
Table 5: Parameters of three different AS-CUT/SDE wafer texturing processes using
chemical solutions to observe effect of IPA concentration
Process
Name
IPA
(vol%)/
(ml)
KOH
(wt%)
Temperature
(oC)
Agitation
rate (rpm)
Process
Time
(min)
1I 2.67/40 3.33 70 125 15,25
2I 5.4/80 3.33 70 125 15,25
3I 8/120 3.33 70 125 15,25
Figure 66 shows the reflection measurements of wafers textured with
parameters given in Table 5. As can be seen from the figure, as the IPA
concentration increases, reflection also increases for 15 min and 25 min process
durations. In other words, area with pyramids decreases. Figure 67 shows mass loss
of wafers. It can be seen from the figure that increasing IPA decreases the etch rate
and mass loss with respective process durations.
300 400 500 600 700 800 900 1000 1100 120020
25
30
35
40
45
50
55
60
40ml IPA
80ml IPA
120ml IPA
R
efl
ec
tio
n (
%)
Wavelength (nm)
(a)
300 400 500 600 700 800 900 1000 1100 1200
15
20
25
30
35
40
45
40ml IPA
80ml IPA
120ml IPA
Re
fle
cti
on
(%
)
Wavelenght (nm)
(b)
Figure 66: Reflection values of as-cut wafers wafers textured with a solution consisting
of 3.33wt % KOH, different IPA concentrations a) 15min and b) 25 min at 70oC with a
speed of 125 rpm
76
15 256
8
10
12
14
16
18
20
Ma
ss
lo
ss
(%
)
Time (min)
40ml IPA
80ml IPA
120ml IPA
Figure 67: Mass loss with respect to time and changing IPA concentration (Lines are
just added to guide the eyes)
Although minimum reflection values were obtained by using 40ml IPA, the
following experiments were not done with this amount. We observed that the surface
of the wafer after texturing process has some features as shown in Figure 68. This
may be due to low IPA concentration causing less wettability of the surface, and H2
bubbles generated during the etching process may not leave the surface as quickly as
possible. That is why these features are not seen the wafers textured with a solution
consisting of 80 ml and 120 ml IPA. Therefore, the concentration of IPA was kept as
5.4 vol% for other experiments.
Figure 68: Textured wafer with a solution consisting of 40ml IPA, 3.33wt % KOH, 15
min at 70oC
77
5.2 Texturing Process with Ultrasonic Agitation
To observe the ultrasonic effect, the caustic etching solutions were prepared
as 3.33 wt% KOH, 5.4 vol% IPA at 60oC and 70
oC for process durations 15 min, 25
min, 35min, and 45 min.
Firstly, 5x5 cm2, p-type, 1-3 Ω-cm, (100) CZ-Si wafers wafers were textured.
The solution was prepared in the glass beaker shown in Figure 33 and stirred by
using magnetic agitation and heated up to 70oC. The heated solution was put in big
ultrasonic cleaner which was filled by water and heated to 70oC to make the solution
temperature stable. The as-cut wafers were textured for process durations 15min and
25 min like process 70K or 70KC.
Figure 69 shows a) reflection values, b) and c) AM 1.5-weighted reflection
(AM 1.5-WR) values. When the reflection values and AM 1.5-WR values are
compared, there is a dramatic fall in the reflection of textured wafers for process
durations 15min and 25 min about 5%, 10% and 20% when compared with process
without cap, process 25min , and process 15min with cap respectively, as seen in
Figure 69 (a). In Figure 69 (c), the AM 1.5-WR values are almost same except for
processes 70M_45min and 80M_45min which were not textured well. Therefore, the
ultrasonic agitation provided low reflection values, which were obtained for process
duration 45min at process temperature at 75oC or 80
oC with magnetic agitation, for
process durations 15min and 25min at 70oC.
78
300 400 500 600 700 800 900 1000 1100 1200
10
20
30
40
50
60
R
efl
ec
tio
n (
%)
Wavelength (nm)
MA with cap_15 min
MA with cap_25 min
MA without cap_15 min
MA without cap_25 min
UA_15 min
UA_15 min
non-texture^2
(a)
14 16 18 20 22 24 26
10
15
20
25
30
35
40
AM
1.5
-we
igh
ted
re
fle
cti
on
(%
)
Time (min)
non-textured^2
MA with cap
MA without cap
UA
(b)
10
12
14
16
18
20
22
A
M 1
.5-w
eig
hte
d r
efl
ec
tio
n (
%)
UA_15min
UA_25min
75M_45min
80M_45min
70M_45 min
75K_35min
70K_45min
80K_45min
(c)
Figure 69: (a) Reflection values for textured as-cut wafers with magnetic agitation
(MA) and ultrasonic agitation (UA) (b) AM 1.5-weighted reflection (AM 1.5-WR)
values (c) Lowest reflection values obtained up to now(Lines are just added to guide the
eyes)
The surface of as-cut wafers textured with ultrasonic agitation was fully
covered by pyramids for process durations 15min, and 25min. Therefore the SEM
pictures were not put here. Figure 70 shows the SEM images in cross sectional view
of wafers whose reflection values are lowest ones. As can be seen, the pyramid
heights are up to 17 µm for wafers textured with magnetic agitation, and the pyramid
heights are up to 4 µm for wafers textured with ultrasonic agitation. Therefore, the
79
pyramid density of wafers textured with ultrasonic agitation is larger about 4 times
than that of wafers textured with magnetic agitation.
Figure 70: SEM images of a) 75M_45min as-cut wafer b) 80K_45min SDE wafer
c) UA_15min as-cut wafer
After that point, ultrasonic texturing processes for process durations 15min
and 25min were repeated, and reproducibility of the process was provided. Then,
15.6x15.6 cm2, p-type, 1-3 Ω-cm, (100) CZ-Si, 25 wafers for each process durations
were textured with parameters given in Table 6, by using a cassette. These wafers
were used for solar cell production.
Table 6: Parameters of two different AS-CUT wafer texturing processes using chemical
solutions
Process
Name
IPA
(vol%)/ ml
KOH
(wt%) /
g
Temperature
(oC)
Agitation Process
Time (min)
70UA 5.4/ 427 3.33/266 70 Ultrasonic 15,25,35,45
60UA 5.4/427 3.33/266 60 Ultrasonic 15,25,35,45
80
Figure 71 shows the reflection measurements and AM 1.5 weighted ones. At
60oC, reflection values decreases sharply from 15 min to 25 min, and 25 min to 35
min. But reflection values decrease slightly at 70oC as texturing time increases.
300 400 500 600 700 800 900 1000 1100 12005
10
15
20
25
30
15 min
25 min
35 min
45 min
non-textured^2
Re
fle
cti
on
(%
)
Wavelength (nm)
(a)
300 400 500 600 700 800 900 1000 1100 12005
10
15
20
25
30
35
40
Re
fle
cti
on
(%
)
Wavelength (nm)
15 min
25 min
35 min
45 min
non-textured^2
(b)
15 20 25 30 35 40 45
12
14
16
18
20
22
24
AM
1.5
We
igh
ted
Re
f le
cti
on
(%
)
Time (min)
70 C
60 C
(c)
Figure 71: Reflection measurements of as-cut wafers textured by ultrasonic agitation at
a) 70oC b) 60
oC c) AM 1.5 weighted reflection values (Lines are just added to guide the
eyes)
Figure 72 a) shows the surface of wafer textured for 15 min at 70oC. The
coverage is inhomogeneous with small pyramids between 0.8-1.5 µm and large
pyramids between 3-6µm. Although surface is covered fully, the reflection values is
high (see Figure 71) because the pyramids have round tips and corners.
81
Figure 72 a) shows the SEM image of wafer textured for 25 min at 70oC. The
coverage almost homogenous with pyramids heights between 4-7 µm. Due to the
homogeneity of pyramids and completed pyramids, the reflection value lower than
that of wafer textured for 15 min about 1% (see Figure 71)
Figure 72 a) shows the SEM images of wafer textured for 35 min and 45 min
at 70oC. The distribution of pyramids is almost same and inhomogeneous. There are
almost same in amount of large pyramids (4-6 µm) and small pyramids (1.5-3
µm).The reflection values almost same as seen in Figure 71.
Figure 72 b) shows the SEM images of wafer textured for 15min, 25min,
35min and 45min at 60oC. For 15min and 25 min, the surface is covered with
pyramids but some of them are not completed, and this makes reflection values
higher than that of wafers textured at 70oC (see Figure 71). But the coverage of 25
min process is higher than that of 15 min process. For 35 min and 45 min, surface is
fully covered with pyramids. The reflection values are almost same.
(a)
Figure 72: SEM images of as-cut wafers textured by ultrasonic agitation at a) 70oC b)
60oC
82
(b)
Figure 72: (Continued)
Figure 73 shows a) the average pyramid height, b) pyramid density with
respect to time and temperature. The average pyramid heights or pyramid densities of
wafers textured for 15 min and 25 min at 60oC are not added because the surface of
them is not fully covered (see Figure 72(b)). The pyramid height of 25 min process
duration is the highest value at 70oC. While the density increases from 35 min to 45
min process time at 70oC, this value decreases at 60
oC.
Figure 73 c) shows the mass loss with respect to time at both process
temperatures. The mass loss sharply increases from 15 min to 25 min process
durations at 70oC and 60
oC. After that point, the amount of etched wafer increases
again but the increment of the mass loss decreases as 2% from 25 min to 35 min,
0.6% from 35min to 45 min at 70oC. It shows that number of (111) planes increases
and the etch rate decreases. The mass loss at 60oC is lower than the mass loss at
70oC about 2%, and the reflection values of wafers textured for 35 min and 45 min at
70 and 60
oC are almost same (see Figure 71 ). In other words, the lowest reflection
83
values can be obtained by etching low amount of wafer (about 7%) at 60oC for 35
min process duration.
15 20 25 30 35 40 450.0
1.0x10-2
2.0x10-2
3.0x10-2
4.0x10-2
5.0x10-2
6.0x10-2
7.0x10-2
8.0x10-2
Py
ram
id D
en
sit
y (m
2)
Time (min)
70o C
60o C
(a)
15 20 25 30 35 40 45
2.5
3.0
3.5
4.0
4.5
Av
era
ge
Py
ram
id H
eig
ht
(m
)
Time (min)
70oC
60oC
(b)
10 20 30 40 503
4
5
6
7
8
9
10
11
Ma
ss
Lo
ss
(%
)
Process duration (min)
70oC
60oC
(c)
Figure 73: a) Pyramid density b) Average Pyramid height c) Mass loss for as-cut
wafers textured with ultrasonic agitation at 70 and 60oC(Lines are just added to guide
the eyes)
5.3 Simulation of Light Trapping
To simulate Si wafer light trapping, we used an optical system design (OSD)
program (Zemax) and a three dimensional solid element modeling program
(Solidworks).
We modeled the wafer textured and non- textured, and put it into the OSD
program. Before making simulations, the absorption and refractive index of mono-Si
data is implemented to produce a glass type acting like mono-Si into the OSD
84
program. The data is taken from PVCDROM and Table 7 shows these values from
250 nm to 1100 nm [41].
Table 7: Absorption coefficient and refractive index values of mono-Si
Wavelength (nm)
Absorption Coefficient (cm
-1)
n (Refractive
Index) Wavelength (nm)
Absorption Coefficient (cm
-1)
n (Refractive
Index)
350 1.04E+06 5.483 730 1.54E+03 3.741
360 1.02E+06 6.014 740 1.42E+03 3.732
370 6.97E+05 6.863 750 1.30E+03 3.723
380 2.93E+05 6.548 760 1.19E+03 3.714
390 1.50E+05 5.976 770 1.10E+03 3.705
400 9.52E+04 5.587 780 1.01E+03 3.696
410 6.74E+04 5.305 790 9.28E+02 3.688
420 5.00E+04 5.091 800 8.50E+02 3.681
430 3.92E+04 4.925 810 7.75E+02 3.674
440 3.11E+04 4.793 820 7.07E+02 3.668
450 2.55E+04 4.676 830 6.47E+02 3.662
460 2.10E+04 4.577 840 5.91E+02 3.656
470 1.72E+04 4.491 850 5.35E+02 3.65
480 1.48E+04 4.416 860 4.80E+02 3.644
490 1.27E+04 4.348 870 4.32E+02 3.638
500 1.11E+04 4.293 880 3.83E+02 3.632
510 9.70E+03 4.239 890 3.43E+02 3.626
520 8.80E+03 4.192 900 3.06E+02 3.62
530 7.85E+03 4.15 910 2.72E+02 3.614
540 7.05E+03 4.11 920 2.40E+02 3.608
550 6.39E+03 4.077 930 2.10E+02 3.602
560 5.78E+03 4.044 940 1.83E+02 3.597
570 5.32E+03 4.015 950 1.57E+02 3.592
580 4.88E+03 3.986 960 1.34E+02 3.587
590 4.49E+03 3.962 970 1.14E+02 3.582
600 4.14E+03 3.939 980 9.59E+01 3.578
610 3.81E+03 3.916 990 7.92E+01 3.574
620 3.52E+03 3.895 1000 6.40E+01 3.57
630 3.27E+03 3.879 1010 5.11E+01 3.566
640 3.04E+03 3.861 1020 3.99E+01 3.563
650 2.81E+03 3.844 1030 3.02E+01 3.56
660 2.58E+03 3.83 1040 2.26E+01 3.557
660 2.58E+03 3.83 1050 1.63E+01 3.554
670 2.38E+03 3.815 1060 1.11E+01 3.551
680 2.21E+03 3.8 1070 8.00E+00 3.548
690 2.05E+03 3.787 1080 6.20E+00 3.546
700 1.90E+03 3.774 1090 4.70E+00 3.544
710 1.77E+03 3.762 1100 3.50E+00 3.541
720 1.66E+03 3.751
85
5.3.1 Simulation of SDE wafer
Figure 74 shows the model of simulation. In this model, there are two
rectangle detectors in front and back of the wafer, one rectangle source and a SDE
wafer. By using this model, reflection detected light intensity by front detector,
transmission detected light intensity by back detector, and absorption, which is
obtained by subtracting addition of reflection and transmission from one hundred,
can be obtained. The wafer thickness is taken as 160 µm because wafers are etched
about 10 µm per side to remove the saw damages, and the wafers used in our
laboratory have thickness about 180 µm.
Figure 74: Model of SDE wafer reflection simulation
Figure 75 a) shows the measured reflection value and simulation result of
SDE wafer. As can be seen, the reflection values of simulation is different from
measured ones about 9% for 350 nm and 375 nm, about 2% between 400 nm and
1025 nm, and the difference is about 4% between 1050 nm and 1100 nm. Our
measurements have an error about 1 %, so the values of simulation obtained between
400nm and 1025 nm differs about 1% from measured reflection values.
Figure 75 shows a) reflection, b) transmission, and c) absorption of light for
wafer thickness of 2, 4, 8, 20, 100, and 180 µm. The reflection, transmission and
absorption values of 350 nm and 500 nm are constant as the thickness increases. It
86
means that for these wavelengths wafer thickness equal and higher than 2 µm is
enough for absorption of light entering the wafer.
300 400 500 600 700 800 900 1000 110030
35
40
45
50
55
60
Experimental
Simulation
Re
fle
cti
on
(%
)
Wavelength (nm )
37.93
37.17
(a)
1 2 4 8 16 32 64 128 25630
35
40
45
50
Re
fle
cti
on
(%
)Wafer thickness (m]
350 nm
500 nm
700 nm
825 nm
1050 nm
1100 nm
(b)
1 2 4 8 16 32 64 128 256
0
10
20
30
40
50
Tra
ns
mis
ion
(%
)
Wafer thickness m)
350 nm
500 nm
700 nm
825 nm
1050 nm
1100 nm
(c)
1 2 4 8 16 32 64 128 256
0
10
20
30
40
50
60
70
Ab
so
rpti
on
(%
)
Wafer thickness (m)
350 nm
500 nm
700 nm
825 nm
1050 nm
1100 nm
(d)
Figure 75: a) Reflection data b) Reflection c) Transmission d) Absorption with respect
to changing thickness (Lines are just added to guide the eyes)
For wavelengths 700 nm and so on, the reflection values decreases as the
thickness increases. It shows that reflection is not due to front reflection, but also due
to light reflected from back surface of the wafer. To obtain zero transmission, the
thickness must be at least 20 µm for 700 nm, 100 µm for 825 nm, and the thickness
must be higher than 2 mm and 8 mm, which are not shown, for 1050 and 1100 nm
respectively.
5.3.2 Simulation of Wafer with Uniformly Distributed Pyramids
First of all, the aim was to simulate random pyramids but we faced with a
problem. The problem is that when one pyramid is put on a rectangle volume, the
87
OSD program perceives as there are a pyramid, air, and a rectangle volume.
Therefore, light passing through the pyramid is reflected back without entering the
rectangle volume. The situation is illustrated in Figure 76 (a). Figure 76 (b) shows
the working model.
(a)
(b)
Figure 76: Simulation of textured wafer obtained by a) assembling one pyramid on a
rectangle volume b) cutting the a part of rectangle volume
The problem was not solved. Therefore, we modeled a wafer surface with
uniformly distributed pyramids as shown in Figure 77 (a). The simulation model is
shown in Figure 77 (b). Here, there is a uniformly textured wafer model instead of
SDE wafer model (see Figure 74) .
(a)
(b)
Figure 77: a) Uniformly textured wafer model b) Simulation model
In the simulations, the total thickness is kept as constant, and it is equal to
sum of heights of front and back pyramids, bare wafer thickness as shown in Figure
77 (b). Figure 78 shows reflection values of reference wafer (randomly textured with
88
average pyramid height 7.5 µm), and model with 1, 6, 10 µm pyramids. From the
figure, the AM 1.5-weighted reflection values of models are higher than reference
about 1%. Although Zemax does calculations by using geometric optic, there are
small differences for wavelength higher than 1050 nm as shown in figure. This is due
to the fluctuations in reflected light from back surface of the wafer for these
wavelengths.
300 400 500 600 700 800 900 1000 11008
12
16
20
24
28
Model with 1m pyramids
Model with 6m pyramids
Model with 10m pyramids
Referans measured
Re
fle
cti
on
(%
)
Wavelength (nm)
(a)
0
2
4
6
8
10
12
14
Model with 1m pyramids
Model with 6m pyramids
Model with 10m pyramids
Referans measured
AM
1.5
-We
igh
ted
Re
fle
cit
on
(%
)
(b)
Figure 78: a)Reflection values for model with 1,6,10 µm pyramids b) AM 1.5-Weighted
Reflection
In addition, two side textured wafers with 6 µm pyramid heights and different
total thicknesses were simulated. The absorption result is shown in Figure 79. From
the figure, the absorption decrease with decrease in thickness between 925 nm and
1100 nm, and the absorption values are same between 350 nm and 925 nm.
89
800 850 900 950 1000 1050 1100
20
30
40
50
60
70
80
90
100 m thick
140 m thick
180 m thickAb
so
rpti
on
(%
)
Wavelength (nm)
Figure 79: Absorption graph of two side textured wafer with 6 µm pyramid height
We also simulated one side textured wafers with different total thicknesses
and different pyramid heights. Thicknesses higher than 180 µm were not taken,
because it will be not cost effective and nowadays trend is production of solar cells
with thin wafers.
Table 8: Parameters of models
Model Name Total Thickness (µm) Pyramid Heights (µm)
100_1side 100 1-6-10
140_1sdie 140 1-6-10
180_1side 180 1-6-10
Figure 80 shows the absorption values for wafers with different pyramid sizes
and thicknesses. The wavelength is taken from 950nm to 1100nm, because the
absorption values almost same up to 950 nm.
In Figure 80 (a), one side textured wafers have higher absorption than two
side textured. Because total reflection probability of light passing through the wafer
and facing with flat surface is higher than that of light facing with surface with
pyramids as shown in Figure 77 (b) Figure 81.
90
As can be seen from Figure 80 (b)- (c)- (d), increasing thickness increases the
absorption for the same pyramid sizes. When thickness is constant and the pyramid
sizes are changed, the absorption values increases with increasing pyramid sizes after
1025 nm.
950 1000 1050 110015
30
45
60
75
90
180_1side_1m pyramids
180_1side_6m pyramids
180_1side_10m pyramids
180_2side_1m pyramids
180_2side_6m pyramids
180_2side_10m pyramids
Ab
so
rpti
on
(%
)
Wavelength (nm)
(a)
950 975 1000 1025 1050 1075 110020
30
40
50
60
70
80
90
Ab
so
rpti
on
(%
)
Wavelength (nm)
180_1side_1 m pyramids
140_1side_1 m pyramids
100_1side_1 m pyramids
(b)
950 975 1000 1025 1050 1075 110020
30
40
50
60
70
80
90
Ab
so
rpti
on
(%
)
Wavelength (nm)
180_1side_6 m pyramids
140_1side_6 m pyramids
100_1side_6 m pyramids
(c)
950 975 1000 1025 1050 1075 1100
30
40
50
60
70
80
90
Ab
so
rpti
on
(%
)
Wavelength (nm)
180_1side_10 m pyramids
140_1side_10 m pyramids
100_1side_10 m pyramids
(d)
Figure 80: Absorption of a) one side and two side textured wafer, b)one side wafer with
1 µm pyramids, c) one side wafer with 6 µm pyramids d) one side wafer with 10 µm
pyramids
Figure 81: Model of a wafer with one side texturing
91
5.4 Solar Cell Results
Screen printed Cz silicon solar cells with dimensions 15.6x15.6 cm2 were
produced. The wafers of cells were 70UA and 60UA whose parameters were given
in Table 9. For each texture parameter, 10 wafers were used (total 90 wafers) to
produce solar cells. To get optimum SiNx layer, each set was divided into groups
with 5 wafers, and average values were taken from one of these groups. The
reference cell was the standard cell produced in GÜNAM laboratory.
Table 9: I-V measurement results by Solar Simulator
Texture
Temperature
(oC)
Time of
texturing
(min)
Average
Jsc
(mA)
Average
Voc
(V)
Average
FF
(%)
Average
Efficiency
(%)
Average
Series
(mΩ)
70UA
70 15 34.612 0.623 79.715 17.204 6.91
70 25 34.648 0.623 79.717 17.218 6.89
70 35 35.686 0.623 78.996 17.587 7.17
70 45 35.687 0.623 79.051 17.601 7.16
60UA
60 15 33.651 0.62 79.535 16.734 7.14
60 25 34.71 0.625 79.546 17.264 7.01
60 35 34.453 0.622 79.744 17.235 6.91
60 45 35.617 0.623 79.314 17.608 6.94
Reference 75 45 34.434 0.619 79.659 16.989 6.84
Table 9 shows the I-V measurement results by Solar Simulator. Figure 82
shows the related curves. Average Jsc increases as process duration increases at
different temperatures. And the Jsc of reference is lower than the average Jsc of the
cells textured with ultrasonic agitation. In Figure 82 (b), reference has the lowest Voc
and the cells textured for 10 min at 70oC have the highest value about 626 mV. The
others have almost same average Voc values about 623 mV except for 15 min
texturing at 60oC. In Figure 82 (c), fill factor values are about 79%. Figure 82 (d)
shows all the efficiencies with respect to all Jsc values of 100 cells. As can be seen,
efficiency is almost linearly proportional to Jsc, and the highest efficiency is 17.737%
92
of cell textured for 45min at 60oC and the related I-V curve is shown in Figure 82 (f).
Average cell efficiency of reference is the lowest one when compared with the others
except for cell textured for 15 min at 60oC as shown in the figure (e). Average cell
efficiency increases with increase in process duration at different process
temperature.
15 20 25 30 35 40 45
33.5
34.0
34.5
35.0
35.5
36.0
70o C
60o C
reference
Texturing time (min)
Av
era
ge
Js
c (
mA
/cm
2)
(a)
15 20 25 30 35 40 45
619
620
621
622
623
624
625
626
Texturing time (min)
70o C
60o C
reference
Av
era
ge
Vo
c (
mV
)
(b)
15 20 25 30 35 40 45
78.9
79.2
79.5
79.8
70o C
60o C
referans
Texturing time (min)
Av
era
ge
Fil
l F
ac
tor
(%)
(c)
31 32 33 34 35 36
15.5
16.0
16.5
17.0
17.5
18.0
Jsc (mA/cm2)
Eff
icie
nc
y (
%)
(d)
Figure 83: Graphs of Average a) Jsc b) Voc c) Fill factor e) Efficiency with respect to
texturing time d) Efficiency- Jsc, f) I-V curve of best cell
93
15 20 25 30 35 40 4516.6
16.8
17.0
17.2
17.4
17.6
17.8
Texturing time (min)
70o C
60o C
referenceA
ve
rag
e E
ffic
ien
cy
(%
)
(e)
0 100 200 300 400 500 600 7000
5
10
15
20
25
30
35
40
- Jsc= 35.85 mA cm-2 / 34.43 mA cm
-2
- Voc= 0.623 mV / 0.619 mV
- FF= 79.43 % / 79.718 %
- Eff= 17.74% / 17.00 %
Js
c (
mA
cm
-2)
Voltage (mV)
Best Reference
Best Cell
(f)
Figure 82: (Continued)
Figure 83 shows the mass loss and the Jsc values with changing texturing
time. As the mass loss increases with increase in process duration, Jsc increases. One
can expected that Jsc of thick wafer should be higher than that of thin wafer for same
reflection values for wafers without ARC. Because the absorption increases when
thickness is increased (see Figure 79 ). In our case, wafers textured for 35 min and
45 min at 60oC are thicker than wafers textured for 35 min and 45 min at 70
oC. But
the Jsc values are almost same. It is because the mass loss is too low to observe such a
change in Jsc. In Figure 79, mass loss about 22% (i.e. 140 µm thickness) decreases
the absorption about 2%, but in our case maximum mass loss is about 10%.
15 20 25 30 35 40 453
4
5
6
7
8
Mass loss
Average Jsc
Process Duration (min)
Ma
ss
lo
ss
(%
)
33.5
34.0
34.5
35.0
35.5
36.0
Av
era
ge
Js
c (m
A/c
m2
)
(a)
15 20 25 30 35 40 454
5
6
7
8
9
10
11
Mass loss
Average Jsc
Process Duration (min)
Ma
ss
lo
ss
(%
)
34.6
34.8
35.0
35.2
35.4
35.6
35.8
Av
era
ge
Js
c (m
A/c
m2
)
(b)
Figure 83: Mass loss and Jsc with changing process durations a) at 60oC b) at 70
oC
94
Figure 84 shows how the average Jsc changes with the AM 1.5 weighted
reflection of wafers with ARC and texturing time. As the reflection decreases, the Jsc
values increases except for 35 min at 60oC. This set shows also the unexpected value
of efficiency, so there should be a production error for this set.
15 20 25 30 35 40 45
3.0
3.5
4.0
4.5
5.0
5.5
6.0
A.M 1.5
Average Jsc
Texturing time (min)
AM
1.5
-we
igh
ted
re
fle
cti
on
(%
)
33.5
34.0
34.5
35.0
35.5
36.0
Av
era
ge
Js
c (m
A / c
m2)
(a)
15 20 25 30 35 40 45
3.0
3.5
4.0
4.5
5.0 A.M 1.5
Average Jsc
Texturing time (min)
AM
1.5
-we
igh
ted
re
fle
cti
on
(%
)
34.5
35.0
35.5
36.0
Av
era
ge
Js
c (m
A / c
m2)
(b)
Figure 84: AM 1.5-weighted reflection and Average Jsc values with respect to changing
texturing time a) at 60oC b) 70
oC
To see the performance of produced cells without effects of series resistance,
the cell parameters were measured by Suns-Voc system. The measurement results
are shown in Table 10.
Table 10: Suns-Voc measurement results
Texture
Temperature
(oC)
Time of
texturing
(min)
Average
Jsc
(mA)
Average
Voc
(V)
Average
FF
(%)
Average
Eff
(%)
70UA
70 15 34.612 0.615 83.113 17.694
70 25 34.648 0.614 83.087 17.700
70 35 35.686 0.614 83.101 18.222
70 45 35.687 0.615 82.94 18.232
60UA
60 15 33.651 0.614 82.587 17.071
60 25 34.71 0.617 82.891 17.767
60 35 34.453 0.616 82.804 17.582
60 45 35.617 0.614 82.864 18.132
Reference 75 45 34.434 0.609 83.013 17.42
95
Figure 85 shows the pseudo efficiencies and fill factors. Pseudo efficiencies
and fill factors are higher than real ones. Real efficiencies and FF are cell parameters
depending on series resistance, and they decrease with increase in series resistances.
15 20 25 30 35 40 45
16.8
17.0
17.2
17.4
17.6
17.8
18.0
18.2
Eff
icie
nc
y (
%)
Time of texturing (min)
pseudo_60
real_60
(a)
15 20 25 30 35 40 45
17.2
17.3
17.4
17.5
17.6
17.7
17.8
17.9
18.0
18.1
18.2
18.3
Eff
icie
nc
y (
%)
Time of texturing (min)
pseudo_70
real_70
(b)
15 20 25 30 35 40 45
79.5
80.0
80.5
81.0
81.5
82.0
82.5
83.0
Av
era
ge
FF
(%
)
Time of texturing (min)
pseudo_60
real_60
(c)
15 20 25 30 35 40 45
79.0
79.5
80.0
80.5
81.0
81.5
82.0
82.5
83.0A
ve
rag
e F
F (
%)
Time of texturing (min)
pseudo_70
real_70
(d)
Figure 85: Pseudo and real efficiencies and fill factors a),c) at 60oC b),d) at 70
oC
process temperature
As shown in Table 10, the Voc value of reference is lower than the others
about 5 mV. If we use Equation 4 to find the saturation currents and then use the Isc
of 70UA_45 min to find reference cell Voc, we see that the calculated Voc of
reference increases about 0.9 mV which is much smaller than 4 mV. It shows that
increase in Voc is not only due to the increase in Jsc. To understand the reason, J01
and J02 values were compared.
96
As shown in Figure 86 (a), J01 values are order of 10-12
A/cm2 which is the
normal range and the values almost same except for reference. The J01 of reference is
slightly higher than the others. Thus, reverse saturation current component is due to
the emitter and bulk recombination.
15 20 25 30 35 40 451.55
1.60
1.65
1.70
1.75
1.80
1.85
1.90
1.95
2.00
70UA
60UA
Reference
J0
1 (
pA
/ c
m2)
Time of texturing (min)
(a)
15 20 25 30 35 40 450
50
100
150
200
250
300
J0
2 (
nA
/ c
m2)
Time of texturing (min)
70UA
60UA
Reference
(b)
Figure 86: a) J01 b) J02 with respect to texturing time
Figure 86 (b) shows J02 values which are order of 10-10
A/cm2
for wafers
60UA 25min and 35min, 70UA 15min, and reference. For the others, they are in
order of 10-9
A/cm2. The values of J02 higher than 10
-10 A/cm
2 shows possible
junction related problems [42].
We can than conclude that low Voc values for the reference sample is
possibly due to the higher recombination of electron hole pairs at the front surface
and low junction quality that leads to higher reverse saturation current.
To observe whether there are high series resistance due to contacts or not,
electroluminescence images like shown in Figure 87 of produced cells were taken.
The all images show almost same characteristics, so the others are not added here.
All images have dots which are due to the contact point of transporters in firing
furnace. There is only dead space at the edges of cells which are cut during edge
isolation.
97
(a)
(b)
Figure 87: Electroluminescence images of a) Reference b) 70UA 45min solar cell
Finally, external quantum efficiencies (EQEs) were measured to observe the
response of cells to different wavelengths. EQE is proportional to passivation and
reflection for short wavelengths, and it is proportional to light trapping for long
wavelengths in general.
Figure 88 a) and b) show the EQE and reflection results of cells with highest
efficiencies and reference. For 45 min texturing process at 60oC, the EQE is higher
than that of reference cell between 350 nm and 700 nm, while reflection is lower
than that of reference. Although the reflection value is slightly higher than that of
reference beyond 700 nm, the EQE is higher than that of reference between 850 nm
and 1050 nm. For 45 min texturing process at 70oC, the situation is similar. Between
350nm and 700nm, the reflection value of 70UA_45min cell is lower than reference;
EQE is higher than that of reference. And EQE is higher than that of reference
between 750 nm and 1100 nm, as the reflection is also higher than that of reference.
These results show that the blue response and the light trapping are higher than those
of reference cell.
98
400 500 600 700 800 900 1000 1100
0
10
20
30
40
50
60
70
80
90
100
E
QE
an
d R
efl
ec
tio
n (
%)
Wavelength (nm)
Reference_EQE
Reference_Reflection
60UA_45min_EQE
60UA_45min_Reflection
(a)
400 500 600 700 800 900 1000 1100
0
10
20
30
40
50
60
70
80
90
100
E.Q
.E a
nd
Re
fle
cti
on
(%
)
Wavelength (nm)
Reference_EQE
Reference_Reflection
70UA_45min_EQE
70UA_45min_Reflection
(b)
400 500 600 700 800 900 1000 1100
30
40
50
60
70
80
90
100
110
I.Q
.E (
%)
Wavelength (nm)
Reference_IQE
60UA_45min_IQE
(c)
400 500 600 700 800 900 1000 1100
30
40
50
60
70
80
90
100
I.Q
.E (
%)
Wavelength (nm)
Reference_IQE
70UA_45min_IQE
(d)
Figure 88: EQE and Reflection-IQE of solar cells produced by using wafers textured
for 45 min a)-c) at 60oC and b)-d) at 70
oC
To observe whether the blue response is higher due to only the reflection
value or not, internal quantum efficiency -wavelength (IQE) curves are also added as
shown in Figure 88 c) and d). Because IQE shows the performance of cells by
excluding the reflection losses.
IQE values are higher than that of reference as shown in Figure 88. It can be
said that the passivation quality of wafers with ultrasonic texturing is higher than that
of reference. To understand the reason, the SEM images of reference wafer after
texturing process was taken. Because, although all process steps are optimized for
the reference, the produced cells with wafer that were textured with ultrasonic
agitation shows better performance than the reference cells. If the quality of process
99
steps, junction quality and wafer quality (see Figure 86 for junction quality and bulk
quality) are same for all samples, the only difference will be the surface
morphologies.
Figure 89 shows the SEM images of wafers whose EQE, IQE and reflection
values are given in Figure 88. As can be seen, the surface morphologies of wafers
textured with ultrasonic agitation are more homogenous than the reference cell. The
reference cell has more small pyramids on big pyramids and big pyramids formed by
small pyramids. This inhomogenity of the surface makes difficult to get good
passivation quality which is related to the front surface recombination of the cells.
Figure 89: SEM image of a) 70UA 45min b) 60UA 45 min c) reference wafer after
texturing process
100
101
CHAPTER 6
CONCULUSIONS
Texturing process is the first step of solar cell production. This process is
carried out by using KOH-IPA solution to minimize the optical losses by forming
random pyramids on wafer surface and thus increases the efficiency of the solar cell.
In this study, the process parameters like temperature, time, and concentrations of
KOH & IPA were optimized by using magnetic agitation. Then process temperature
and duration were minimized by using ultrasonic agitation. Major conclusions of our
studies are summarized below.
The results related with magnetic agitation are;
1. As the process duration increases, covered area and pyramid sizes increases for
magnetic agitation. But it is not enough to get fully covered surfaces by just
increasing the time. All process parameters must be optimized.
2. For process parameters 4.2 wt% KOH, 5.4vol% IPA, as-cut and SDE wafers had
lowest reflection values for 45 min process duration at 70, 75 and 80oC. But
when the reflection results for 45 min were compared, lowest value was obtained
at 75oC. The difference between the highest and lowest reflection was 8% and
6% for as-cut and SDE respectively for 45 min process time.
3. For process parameters 3.3 wt% KOH, 5.4vol% IPA, again the lowest reflection
values were obtained for 45 min process time. The difference between the
highest and lowest reflection was 1% and 2% for as-cut and SDE respectively
for 45 min process time. It shows that these process parameters were less
sensitive to temperature when compared with previous one. Wafers textured
with 3.3wt% KOH had lower reflection values than wafers textured with 4.2wt
% KOH for 15, 25, and 35 min texturing time. This shows that decreasing KOH
102
concentration from 4.2 wt% to 3.3 wt% improved the process for short process
times.
4. A cap was used to close the glass beaker in which processes were carried out.
Closing the beaker increases the pressure on the solution and the evaporation of
water and IPA decreases. This makes the concentrations almost constant. For
process parameters 4.2 wt% KOH, 5.4 vol% IPA at 75oC, the wafers textured
for 15 min and 25 min had lower reflection than before. When the process was
carried out for 45 min, the pyramid densities increased for both as-cut and SDE
and this decreased the reflection slightly.
5. Increasing speed and changing IPA concentration did not improve process for
15 min and 25 min very much.
6. There are some problems faced during texturing process with magnetic
agitation.
The reproducibility of the process was hard for some cases in which the
process duration was lower or equal 25 min for all process temperature
especially 70oC.
The surfaces of wafers were not uniformly textured, and there were differences
in reflection between the faces again for process durations lower or equal to 25
min.
These results show that it is hard to minimize process parameters especially
time and temperature by using magnetic agitation. We decided to change agitation
type with ultrasonic agitation.
1. As time increases, reflection decreases again.
2. The wafer surface was fully covered by texturing with 3.3wt% KOH, 5.4 wt%
IPA for 15 min at 70oC and 35 min 60
oC.
3. 15 min texturing had higher reflection due to pyramids with round tips.
4. The pyramid density of ultrasonic agitation was higher than that of magnetic
one about 5 times. Therefore small pyramids were obtained.
5. Having low boiling point (82.4oC), which makes it necessary to re-dose IPA
into the solution due to evaporation, is the main problem of the standard
103
solution and causes consuming high concentrations of IPA and leads to high
production costs, because IPA is expensive. Therefore by using ultrasonic
agitation, cost effective process was obtained by decreasing temperature and
shorting process time.
We did also simulations of SDE, one side and two side uniformly textured wafers.
The results are;
1. The reflection for SDE wafer is not only due to light reflected from front
surface but also back surface for wavelength higher than 700 nm and thickness
lower than 70 m.
2. The absorption decreases for wavelength higher than 850 nm as thickness
decreases for simulation model one side and two side textured wafers whose
thicknesses higher or equal to 100 m.
3. One side textured wafers shows higher absorption than two side textured,
because the total reflection probability of light not coming perpendicular to flat
surface is higher than that of light to surface with pyramids.
The solar cell results of wafers textured with ultrasonic agitation are;
1. The Jsc value was higher than the reference cell (34.4 mA/ cm2)
about 1.25 mA/
cm2 for wafers with efficiency about 17.6%
2. The Voc values were higher than the reference cell (619 mV) about 4 mV for
almost all wafers textured with ultrasonic agitation.
3. The efficiencies were higher than the reference cell (17%) about at least 0.2%
and at most 0.7%. The highest efficiency was 17.74% of wafers textured for 45
min 60oC with ultrasonic agitation.
4. From EQE and IQE results, the wafers textured with ultrasonic agitation had
uniformly distributed pyramids with heights less than 4 µm leading had better
blue response, surface passivation and light trapping.
To sum up, the texturing process with ultrasonic agitation has a potential to
minimize the process parameters. If more experiments are made about this texturing
technique, more cost effective processes will be able to obtain. If the solar cell
104
production parameters are optimize for wafers textured with ultrasonic agitation,
higher efficiency values will be obtained.
105
BIBLIOGRAPHY
[1] “Enviromental & Regulatory Specialists.” [Online]. Available:
http://earsi.com/ggp01.htm. [Accessed: 12-Oct-2014].
[2] “ISE Photovoltaics Report.” [Online]. Available:
https://www.ise.fraunhofer.de/en/downloads-englisch/pdf-files-
englisch/photovoltaics-report-slides.pdf. [Accessed: 10-Nov-2014].
[3] “PVCDROM.” [Online]. Available: http://pveducation.org/pvcdrom/.
[Accessed: 12-Nov-2014].
[4] D. M. Chapin, C. S. Fuller, and G. L. Pearson, “A new silicon p-n junction
photocell for converting solar radiation into electrical power,” J. Appl. Phys.,
vol. 25, pp. 676–677, 1954.
[5] W. Schockley and H. J. Quesisser, “Detailed balance limit of efficiency of pn
junction solar cells,” J. Appl. Phys., vol. 32, pp. 510–519, 1951.
[6] J. Nestor and X. Quiebras, “Wet chemical textures for crystalline silicon solar
cells,” pp. 1–124, 2013.
[7] M. A. Green, “Silicon Solar Cells: State of Art,” Philosophical Transactions
of The Royal Society A, 2013. [Online]. Available:
http://dx.doi.org/10.1098/rsta.2011.0413. [Accessed: 12-Nov-2014].
[8] “Nanosolar CIGS Solar Cells.” [Online]. Available:
http://www.solarfeeds.com/rip-nanosolar/. [Accessed: 01-Jan-2015].
[9] M.S. Thesis,E. H. Çiftpınar, Selective Emitter Formation via Single Step
Doping Through Laser Patterned Mask Oxide Layer for Monocrystalline
Silicon Cells: Middle East Technical University, 2014.
[10] M.S. Thesis, S. H. Sedani, Fabrication and Doping of Thin Crystalline Si
Prepared by E-Beam Evaporation on Glass Substrate: Middle East Technical
University, 2013.
[11] B. M. Kayes, L. Zhang, R. Twist, I. Ding, and G. S. Higashi, “Flexible Thin-
Film Tandem Solar Cells With > 30 % Efficiency,” IEEE J. Photovoltaics,
vol. 4, no. 2, pp. 729–733, 2014.
106
[12] M.S. Thesis, F. Es, Fabrication and Characterization of Single Crystalline
Silicon Solar Cells: Middle East Technical University, 2010.
[13] G. M. Research, “Building Integrated Photovoltaics: An emerging market.”
[Online]. Available: http://www.solarserver.com/solar-magazine/solar-
report/solar-report/building-integrated-photovoltaics-an-emerging-
market.html. [Accessed: 15-Dec-2014].
[14] “NSF / ONR Workshop on Key Scientific and Technological Issues for
Development of Next-Generation Organic Solar Cells.” [Online]. Available:
http://web.utk.edu/~opvwshop/. [Accessed: 01-Jan-2015].
[15] C. S. Solanki, “Solar Photovoltaics: Fundamentals Technologies And
Applications,” PHI Learning Pvt. Ltd., 2009, pp. 323–324.
[16] B. Hoffman, V. Sivakov, S. W. Schmitt, M. Y. Bashouti, M. Latzel, J. Dluhoš,
and J. Jiruse, “Wet – Chemically Etched Silicon Nanowire Solar Cells:
Fabrication and Advanced Characterization,” P. X. Peng, Ed. 2012, pp. 211–
230.
[17] “Series Resistance.” [Online]. Available:
http://pveducation.org/pvcdrom/solar-cell-operation/series-resistance.
[Accessed: 01-Feb-2015].
[18] “Quantum Efficiency.” [Online]. Available:
http://www.pveducation.org/pvcdrom/solar-cell-operation/quantum-efficiency.
[Accessed: 01-Feb-2015].
[19] J. D. Plummer, M. D. Deal, and P. B. Griffin, “Silicon VLSI technology:
fundamentals, practice and modeling,” Prentice Hall, 2000, pp. 618–619.
[20] R. N. Ghoshtagore, “Phosphorus diffusion processes in SiO2 films,” Thin
Solid Films, vol. 25, no. 2, pp. 501–513, 1975.
[21] K. Sakamoto, K. Nishi, F. Ichikawa, and S. Ushio, “Segregation and Transport
Coefficients of Impurities at the Si/SiO2 Interface,” J. Appl. Phys., vol. 61, no.
4, pp. 1553–1555, 1987.
[22] M.S. Thesis, M. Bilgen, Comparison of Crystal Si Solar Cells Fabricated on
P-Type and N-Type Float Zone Silicon Wafers: Middle East Technical
University, 2013.
[23] J. Nelson, The Physics of Solar Cells, 1st ed. London: Imperial College Press,
2003.
107
[24] “Anti-Reflection Coatings.” [Online]. Available:
http://pveducation.org/pvcdrom/design/anti-reflection-coatings. [Accessed:
01-May-2015].
[25] B. Streetman and S. Banerjee, Solid State Electronic Devices, 6th ed. PHI
Learning, 2009.
[26] F. Colville and C. Dunsky, “Lasers Support Emerging Solar Industry Needs,”
Photonic Spectra, 2008. [Online]. Available:
http://www.photonics.com/Article.aspx?AID=33002. [Accessed: 01-Feb-
2015].
[27] “Reference Solar Spectral Irradiance: ASTM G-173.” [Online]. Available:
http://rredc.nrel.gov/solar/spectra/am1.5/astmg173/astmg173.html. [Accessed:
23-Jun-2015].
[28] M. Moreno, D. Murias, J. Martínez, C. Reyes-Betanzo, a. Torres, R.
Ambrosio, P. Rosales, P. Roca i Cabarrocas, and M. Escobar, “A comparative
study of wet and dry texturing processes of c-Si wafers for the fabrication of
solar cells,” Sol. Energy, vol. 101, pp. 182–191, 2014.
[29] H. Seidel, L. Csepregi, A. Heuberger, and H. Baumgärtel, “Anisotropic
etching of crystalline silicon in alkaline solutions,” J. Electrochem. Soc., vol.
137, no. 11, p. 3612, 1990.
[30] D. B. Lee, “Anisotropic etching of silicon,” J. Appl. Phys., vol. 40, no. 11, pp.
4569–4574, 1969.
[31] E. D. Palik, “A raman study of etching silicon in aqueous KOH,” J.
Electrochem. Soc., vol. 130, no. 4, p. 956, 1983.
[32] J. B. Price, “Anisotropic etching of silicon with KOH-H2O-isopropyl
alcohol,” J. Electrochem. Soc., vol. 120, no. 3, p. C96, 1973.
[33] I. Kashkoush, G. Chen, and D. Nemeth, “Characterization of c-Si
Texturization in wet KOH/IPA and its Effect on Cell Efficiency,” pp. 5–8.
[34] A. K. Chu, J. S. Wang, Z. Y. Tsai, and C. K. Lee, “A simple and cost-effective
approach for fabricating pyramids on crystalline silicon wafers,” Sol. Energy
Mater. Sol. Cells, vol. 93, no. 8, pp. 1276–1280, 2009.
[35] L. Wang, F. Wang, X. Zhang, N. Wang, Y. Jiang, Q. Hao, and Y. Zhao,
“Improving efficiency of silicon heterojunction solar cells by surface texturing
of silicon wafers using tetramethylammonium hydroxide,” J. Power Sources,
vol. 268, pp. 619–624, 2014.
108
[36] K. Birmann, M. Demant, S. Rein, and B. Bläsi, “Optical Characterization of
Random Pyramid Texturization,” Small, no. September, pp. 5–9, 2011.
[37] J. D. Hylton, a. R. Burgers, and W. C. Sinke, “Alkaline Etching for
Reflectance Reduction in Multicrystalline Silicon Solar Cells,” J.
Electrochem. Soc., vol. 151, no. 6, p. G408, 2004.
[38] M.S. Thesis, O. Demircioğlu, Optmization of Metallization in Crystalline
Silicon Solar Cells: Middle East Technical University, 2012.
[39] “High Lambertian reflectance over their effective spectral range.” [Online].
Available: https://www.labsphere.com/wp-
content/uploads/2015/05/Spectralon-Diffuse-Reflectance-Standards.pdf.
[Accessed: 13-Dec-2014].
[40] R. A. Sinton, A. Cuevas, and M. Stuckings, “Quasi-steady-state
photoconductance, a new method for solar cell material device and
characterization,” in Proc. 25th Photovoltaic Specialists Conference, 1996, pp.
457–460.
[41] “Optical Properties of Silicon.” [Online]. Available:
http://pveducation.org/pvcdrom/materials/optical-properties-of-silicon.
[Accessed: 01-May-2015].
[42] O. Breitenstein, J. Bauer, P. P. Altermatt, and K. Ramspeck, “Influence of
Defects on Solar Cell Characteristics,” Solid State Phenom., vol. 156–158, pp.
1–10, 2009.